Automatic Release Notes Summarization with Large Language Models
Qianru Meng, Zhaochun Ren, Joost Visser
Leiden University
Release notes are critical documents summarizing changes in software releases, including bug fixes, new features, and performance enhancements. Instead of extractive summarization, which often lacks contextual and semantic understanding, we use generative models to enhance the creation of release notes. Our approach leverages large language models to generate high-quality, contextually rich summaries from commit messages, highlighting the most important changes. This method ensures release notes that effectively serve the diverse needs of stakeholders, not just end-users but also developers and project managers.