Skip to Main Content
Spectrum based fault localization techniques such as Tarantula and Ochiai calculate the suspiciousness score of a program statement using the number of failing and passing test cases that execute the statement. These techniques implicitly assume that all test cases are equally important. However, research on test case generation and selection techniques has shown that using certain test cases can lead to more effective fault localization than others. The proposed research aims to improve the effectiveness of spectrum based fault localization by incorporating the relative importance of different test cases in the calculation of suspiciousness scores.