LLM Contribution Summarization in Software Projects

This full paper in innovative practice provides an automated tool to summarize individual code contributions in project-based courses with external clients. Real industry projects offer valuable learning opportunities by immersing students in authentic problems defined by external clients. However, the open-ended and highly variable scope of these projects makes it challenging for instructors and teaching assistants to provide timely and detailed feedback. This paper addresses the need for an automated and objective approach to evaluate individual contributions within team projects. In this paper, we present a tool that leverages a large language model (LLM) to automatically summarize code contributions extracted from version control repositories. The tool preprocesses and structures repository data, and uses PyDriller to isolate individual contributions. Its uniqueness lies in the combination of LLM prompt engineering with automated repository analysis, thus reducing the manual grading burden while providing regular and informative updates. The tool was assessed over two semesters during a three-week, full-time software development sprint involving 65 students. Weekly summaries were provided to teams, and both student and faculty feedback indicated the tool's overall usefulness in informing grading and guidance. The tool reports, in large proportion, activities that were in fact performed by the student, with some failure to detect students' contribution. The summaries were considered by the instructors as a useful potential tool to keep up with the projects.