A Fault Localization Method Based on Conditional Probability

The fault localization is an active research topic in the field of software engineering, in which the program spectrum-based fault localization(SFL) is an effective method. It's a base for program spectrum-based fault localization to find the internal linkage between program spectrum and executive result. In this paper, through analysis of the internal linkage between program spectrum and executive result, the conception of the conditional probability in statistics is introduced and four models of the conditional probability (p-model) are designed to quantify the relationship between both. Based on p model and combined with the information theory of the low-probability events containing more information, a new method of fault localization is proposed SCP (Suspiciousness based on Conditional Probability). In order to verify the effectiveness of the SCP method, experiments are conducted vith eight public data sets. From the results, it's shown that the SCP method has certain advantages in the fault localization.