一句话：系统分析 280 个 Java 回归缺陷的诱导变更，并评估与改进 LLM 在提交级回归缺陷检测中的表现。
回归缺陷在真实软件项目中十分常见，但已有研究多关注其总体特征，尚缺乏对“诱导回归缺陷的代码变更”的细粒度分析；在 LLM 广泛应用的背景下，能否以及如何从开发提交中提前识别这类缺陷也不清楚。本文对来自开源 Java 项目的 280 个回归缺陷进行系统研究，分析其诱导变更的根本原因及所关联的开发活动，并评估 LLM 检测回归缺陷诱导提交的有效性，解释其成功与失败原因。该工作刻画了回归缺陷诱导变更的特征，并为改进基于 LLM 的回归缺陷检测提供了实证基础。
Regression bugs are common in real-world software projects. Although several studies have characterized various aspects of these bugs, a detailed analysis of the characteristics of code changes that introduce regression bugs is still lacking. It is also not clear whether and how regression bugs can be identified, in the era of wide usage of large language models (LLMs), in code commits during software development in the first place. To fill this gap, our study systematically analyzes 280 regression bugs from open-source Java projects, examining both the root causes and the associated development activities of regression bug-inducing changes, while also evaluating the effectiveness of LLMs in detection such commits and explaining their underlying causes.
@inproceedings{paperbot11727,
title = {Characterizing Regression Bug‑Inducing Changes and Improving LLM‑Based Regression Bug Detection},
author = {Xuezhi Song and Yijian Wu and Bihuan Chen and Zhengjie Lu and Shuning Liu and Xin Peng},
booktitle = {ICSE},
year = {2026},
doi = {10.1145/3744916.3773182}
}