CoDocBench: A Dataset for Code-Documentation Alignment in Software Maintenance
CoDocBench is a dataset designed for research on code-documentation alignment in software maintenance. It consists of 4,573 code-documentation pairs extracted from 200 open-source Python projects. Each data point includes metadata such as version history, unified diffs, docstring changes, code changes, and repository information. Additionally, users can create their own custom datasets by extracting code-documentation pairs from other repositories using the provided tools and scripts.
Institutions
- University of California, Davis (US)
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2024-11-30
- DOI
- https://doi.org/10.5281/zenodo.14251623
- Primary Topic
- Software Engineering Research
- Type
- article
- Field-Weighted Citation Impact
- 0.00