Data model / ontology for scientific reproducibility

[mlieberman85]

Summary

darnit-reproducibility today ships as a compliance-implementation plugin with control checks specific to scientific-computing hermetic-build practices. Complement it with a formal data model / ontology so that reproducibility properties darnit checks can be spoken about, exchanged, and composed with the broader scientific-reproducibility ecosystem's vocabulary.

Sibling to #392 (supply-chain-security GUAC ontology) -- same class of work in a different domain.

Scope (research + design first)

Survey prior art and identify the vocabulary darnit-reproducibility's controls should map onto:

• PROV (W3C provenance vocabulary): the canonical academic ontology for "who / what / when produced this artifact". Widely used in research data pipelines.
• Citation File Format (CFF) + FORCE11 Software Citation Principles: how software artifacts identify themselves and their reproducible-execution recipe.
• BagIt / RO-Crate (Research Object Crate): container formats for bundling data + code + metadata reproducibly.
• RSF (Reproducibility Scorecard Framework, ML Reproducibility Checklist, NeurIPS reproducibility checklist): domain-specific reproducibility checklists that map cleanly to compliance controls.
• REES (Reproducible Environments Executor Service, Binder ecosystem): the closest thing to a hermetic-build-for-research contract.
• Hermetic build vocabulary darnit already uses (Witness attestations per feature 024, GoReleaser / SLSA overlaps with Data model / ontology for supply chain security (GUAC-aligned) #392).

Map darnit-reproducibility's control verdicts and evidence onto the survey. Identify:

• Overlaps we can adopt directly.
• Gaps in existing ontologies that darnit-reproducibility's use cases surface.
• Whether an existing ontology suffices or a small darnit-specific extension is warranted.

Non-goals for the research phase

• Rewriting darnit-reproducibility's existing controls to emit the new shape (that's a follow-up once the ontology is agreed).
• Adjudicating whether "reproducibility" and "supply-chain security" ontologies (this issue vs Data model / ontology for supply chain security (GUAC-aligned) #392) should merge or stay separate. Answer that question in the spec once both surveys exist.

Deliverables

• A docs/research/reproducibility-ontology.md (or a spec-kit spec) with:
  • Survey of the prior art above.
  • darnit-reproducibility control -> ontology-term mapping table.
  • Gap analysis.
  • Recommendation: adopt existing vocabulary (PROV / RO-Crate / etc.), extend one, or emit a small darnit-reproducibility-specific shape.

Constitution alignment

• Data-model / research work; no code paths touched in this issue.

Related

• Data model / ontology for supply chain security (GUAC-aligned) #392 (Supply-chain-security ontology) -- sibling; may converge or stay separate depending on the surveys.
• RE-02.01 HermeticBuild: evolve beyond grep-based heuristic toward verifiable hermeticity #227 (RE-02.01 HermeticBuild: evolve beyond grep-based heuristic) -- consumer of the resulting vocabulary.