Abstract:Study automated localization of software bugs on the basis of an important statistical model-based bug localization,called SOBER.By program research and large numbers of instance analysis,find the SOBER's limitation which will cause localization errors of software bugs because of predicate relativity.Come up with a new solution about evaluation bias of relevant predicates and conduct instance study.The results show that the study preferably solves the problem of predicate interference and greatly improves SoBER-based automated localization of program bugs.

Improvement of Relevant Predicate Evaluation Bias Method for SOBER-based Fault Localization