Selecting test cases for result inspection to support effective fault localization

Abstract

Fault localization techniques help locate faults in source codes by exploiting collected test information and have shown promising results. To precisely locate faults, the techniques require a large number of test cases that sufficiently exercise the executable statements together with the label information of each test case as a failure or a success. However, during the process of software development, developers may not have high-coverage test cases to effectively locate faults. With the test case generation techniques, a large number of test cases without expected outputs can be automatically generated. Whereas the execution results for generated test cases need to be inspected by developers, which brings much manual effort and potentially hampers fault-localization effectiveness. To address this problem, this paper presents a method to select a few test cases from a number of test cases without expected outputs for result inspection, and in the meantime selected test cases can still support effective fault localization. The experimental results show that our approach can significantly reduce the number of test cases that need to be inspected by developers and the effectiveness of fault localization techniques is close to that of whole test cases.