综合利用语句和序列可疑度的软件故障定位方法

Software Fault Localization by Integrating the Suspiciousness of Statements and Sequences

自动故障定位对于提高软件调试的效率有重要意义.本文提出利用语句、数据流和控制流的频谱信息并基于局部最优性生成和计算语句序列的可疑度,并综合语句所在序列的可疑度和语句的可疑度以获得语句的最终可疑度排序.在西门子测试程序集上与两种基于语句可疑度的故障定位方法做了实验性能比较,结果验证了所提出方法的有效性.

Automated fault localization is significant for improving the efficiency of software debugging. We propose to generate state- ment sequences and compute their suspiciousness from the spectrum information of statements as well as control and data dependency relations based on local optimality. Then the suspiciousness of sequences and statements are integrated to produce a suspiciousness ranking of each statement. Performance comparison with two methods based on suspiciousness of statements on the Siemens Suite has validated the effectiveness of proposed approach.