Mining Associations to Improve the Effectiveness of Fault Localization

Abstract

Coverage-based fault localization(CBFL) techniques find the fault-related positions in programs by comparing the execution statistics of passed executions and failed executions have been proven to be efficient by several empirical studies.However,these techniques assess the suspiciousness of program entities individually,whereas the individual coverage information cannot reflect the complicated control-and data-dependency relationships,and thus oversimplify the execution spectra.In this paper,we first use motivating examples to show the impact of the cause-effect relationship on the effectiveness of CBFL.Second,we propose the rules of program failures and design the execution analysis model based on the coverage of different program execution spectrum.By computing the frequency items for statements with high suspiciousness,we also bring out the coverage vector to mine fault-prone statements.The controlled experiments based on the SIR benchmarks indicate that our technique is promising.