Abstract: Software fault localization is an important research field in software engineering, and the Coverage-based Fault Localization（CFL） is an important method among the existing approaches. Coincidental correctness（CC） test cases could have negative effects on the CFL, so that how to find CC test cases or reduce the impact of the CC test cases is significant to improve the CFL effect. By the research on how does CC test cases affect CFL, discover a way to find what kinds of test case are CC test cases which has no false positive situation. Based on this, propose a method to improve CFL on CC. It shows that this method can improve most of the CFL through the experiments.

Key words: fault localization, coincidental correctness, software testing, debugging

摘要： 软件错误定位是目前软件工程领域的重要研究课题，其中基于程序谱的错误定位（CFL）是一类重要的方法。偶然性正确测试用例对于CFL有着很大的负面影响，因此如何发现该类测试用例或者避免它们对CFL的影响对于提高CFL方法的定位效果有着重要的意义。通过分析偶然性正确对CFL定位方法的影响，发现一种没有误判率（false positive）的寻找偶然性正确测试用例的方法。在此基础上，提出了一种基于偶然性正确测试用例发现的CFL定位方法。通过实验表明，该方法可以普遍改善已有CFL方法的定位效果。

关键词: 错误定位, 偶然性正确, 软件测试, 调试