Abstract Automatic fault localization is essential for software engineering. However, fault localization suffers from the interactions among multiple faults. Our previous research revealed that the fault-coupling effect is responsible for the weakened fault localization performance in multiple-fault programs. On the basis of this finding, we propose a Test Case Restoration Method based on the Genetic Algorithm (TRGA) to search potential coupling test cases and conduct a restoration process for eliminating the coupling effect. The major contributions of the current study are as follows: (1) the construction of a fitness function to measure the possibility of failed test cases becoming coupling test cases; (2) the development of a TRGA that searches potential coupling test cases; (3) and an evaluation of the TRGA efficiency across 14 open-source programs, three spectrum-based fault localizations, and two parallel debugging techniques. The results revealed the TRGA outperformed the original fault localization techniques in 74.28% and 78.57% of the scenarios in the best and worst cases, respectively. On average, the percentage improvement was 4.43% for the best case and 2% for the worst case. A detailed discussion of TRGA parameter settings is also provided.

中文翻译：

---

一种基于遗传算法的测试恢复方法，用于多故障程序中的有效故障定位

摘要 自动故障定位对于软件工程是必不可少的。然而，故障定位受到多个故障之间的相互作用的影响。我们之前的研究表明，故障耦合效应是导致多故障程序中故障定位性能减弱的原因。在此基础上，我们提出了一种基于遗传算法（TRGA）的测试用例恢复方法来搜索潜在的耦合测试用例，并进行恢复过程以消除耦合效应。目前研究的主要贡献如下：（1）构建适应度函数来衡量失败的测试用例成为耦合测试用例的可能性；(2) 开发搜索潜在耦合测试用例的 TRGA；(3) 以及对 14 个开源程序、三个基于频谱的故障定位和两个并行调试技术的 TRGA 效率的评估。结果表明，在最佳和最坏情况下，TRGA 分别在 74.28% 和 78.57% 的场景中优于原始故障定位技术。平均而言，最好情况下的百分比改进为 4.43%，最坏情况下为 2%。还提供了对 TRGA 参数设置的详细讨论。