Mutation-Based Fault Localization Based On Test Oracle

With the development of information technology,people are increasingly demanding software quality.There are some vulnerabilities in the software,and the process of finding these vulnerabilities is called fault localization.Fault localization is a critical step in discovering and fixing errors to improve software quality.Among the many fault localization methods,Mutation Based Fault Localization(MBFL)is a method with high positioning accuracy.This method uses a mutated method to simulate the real fault of the software and calculates the similarity to assist the positioning.The drawback of this method is that it needs to perform a mutation operation on the statement to generate large quantity of mutants,and at the same time execute all the test cases for each mutant.This results in a time-consuming resource overhead for the method and is therefore rarely used by industry.In this paper,the method of combining test oracle with MBFL is proposed for the first time.The execution result information of test cases executed with a small number of mutants is used to construct a prediction model,and then the prediction model is used to predict the results of executing test cases for the remaining mutants.Since the time overhead of predicting is much less than the time overhead of executing,the efficiency of the MBFL can be greatly improved.In order to study the influence of this method on the fault localization accuracy,on the one hand,based on the skeptical calculation formula,this paper theoretically analyzes the influence of the prediction model method on the error location accuracy.On the other hand,experimental verification is performed on 99 error version programs in 5 programs of SIR library and 208 error version programs in 5 programs in Defects4J.At the same time,it can reduce the execution time by about 80%and the fault positioning accuracy is not significant.It improves the efficiency of MBFL.