Reproduce “RepoDoc: A Knowledge Graph-Based Framework to Automatic Documentation Generation and Incremental Updates” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Dong Xu, Mingwei Liu, Xiwen Wang, Jianfeng Zhong, Zibin Zheng

RepoDoc, developed at Sun Yat-sen University, introduces a knowledge graph-based framework for automated, multi-level software documentation generation and incremental updates. It achieves superior documentation completeness and quality, significantly reduces generation time and token costs compared to existing methods, and provides efficient, accurate updates in evolving codebases.

Key insights

  • Leveraging a Repository Knowledge Graph (RepoKG) as a semantic backbone enables the generation of modular, cross-referenced documentation that reflects the codebase's true architecture.
  • Precise context retrieval, facilitated by the RepoKG, avoids context flooding in LLMs, leading to significantly reduced token consumption and faster documentation generation times.
  • Semantic Impact Propagation through the RepoKG accurately traces how code changes affect dependencies, allowing for efficient and accurate incremental updates rather than full regeneration.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.