Context-Selection for Git Diff: Budgeted Submodular Maximization on Typed Code Dependency Graphs
Abstract
Formalizing diff context selection as budgeted submodular maximization over typed Code Property Graphs with multi-resolution fragments and information-need-driven utility, where relevance propagates via weighted structural dependencies from edited code using Personalized PageRank. This paper presents the theoretical foundation for the TreeMapper project's context selection algorithm. Empirical evaluation is the subject of forthcoming work. Community testing and bug reports welcome.
Fields of Study & Subjects
Software Engineering Research
Software Engineering Techniques and Practices
Information Retrieval and Search Behavior
Publication details
- Journal / source
- Zenodo (CERN European Organization for Nuclear Research)
- Publication date
- 1 March 2026
- Language
- EN
- Document type
- Preprint