PurPliance

This repository contains the component that extracts privacy-statement tuples from policy sentences of PurPliance.

Run code

Download and extract the NER model

Download NER model en_core_web_lg.high_f1_data_org.model.tar.xz and extract the file to src/oppnlp/analyze/pded/models/:

src/oppnlp/analyze/pded/models $ tar -xvf en_core_web_lg.high_f1_data_org.model.tar.xz

Set up dependencies and run test code

To run the code, use a virtual environment and run the test as follows:

# Create new conda python environment.
conda create -n purpliance_oss python=3.8
conda activate purpliance_oss

# Install the current package.
pip install -e .
pip install -r requirements.txt

# Test: Extract privacy statements from each file in test/policies and output
# as json files in test/policies/stmt.
bash test/test.sh

# Extract privacy statements from each file in the $INPUT_DIR
# json files in $INPUT_DIR/stmt.
# INPUT_DIR contains plain sentencized text files *.txt to be analyzed.
# Each non-blank line in the file should contain one and only 1 sentence.
python src/runner/analyze/priv_stmt/run_priv_stmt_extractor.py $INPUT_DIR

The code was tested on Ubuntu 18.04 and MacOS 12.6.

Publication

Duc Bui, Yuan Yao, Kang G. Shin, Jong-Min Choi, and Junbum Shin.

Consistency Analysis of Data-Usage Purposes in Mobile Apps.

In Proceedings of the 2021 ACM SIGSAC Conference on Computer and Communications Security (CCS '21).

Association for Computing Machinery, New York, NY, USA, 2824–2843.

https://doi.org/10.1145/3460120.3484536

License

PurPliance is licensed under the BSD-3-Clause License (See LICENSE.txt).

Acknowledgement

This repo uses code from PolicyLint/PoliCheck repository.