Research On The Optimization Of MBFL Technology Based On Mutant Bias And Contribution

With the continuous development of modern technology,the application of software has penetrated all fields of society.In the process of software development,faults can not be avoided,so fault localization becomes a key step in the process of software test.Typically,the cost of the software test process consumes 50% to 80% of the total maintenance cost.With the complexity of software scale and structure,the number of faults also increases,and the possible consequences become more and more serious.Therefore,the performance of fault localization technology is increasingly required.An excellent fault localization technique can effectively shorten the time for developers to fix errors and reduce the cost of maintenance.Mutation-based fault localization(MBFL)is one of the most widely studied techniques in fault localization.MBFL uses methods from mutation analysis to generate mutants to find potential bugs in the program.A large number of existing studies have proposed optimization methods for MBFL in terms of localization efficiency and accuracy.However,our study found that these methods ignore the difference between mutants on the correct entity and the wrong entity,which we call mutant bias.In this study,we analyzed and determined the effect of mutant bias on MBFL.It was found that the mutant bias affects the suspiciousness of statements and negatively affects the localization accuracy of MBFL.To mitigate mutant bias,we propose Delta for Mutants(Delta4Ms),a model that captures mutant bias from mutants of the same statement.Then the true suspiciousness is obtained by eliminating the mutant bias.In addition,we also propose a contribution-based test case reduction strategy Contribution-based Test Case Reduction strategy(CTCR)for MBFL,which can reduce about 80% of test cases while maintaining the localization accuracy of MBFL,thus greatly reducing the localization cost of MBFL.In addition,we combine CTCR with the Delta4 Ms technique and propose the C-Delta4 Ms technique to further improve localization efficiency and accuracy.To evaluate the performance of the proposed methods,we conduct detailed experiments on 317 real faulty programs of Codeflaws.The experimental results show that Contribution-based Delta for Mutants(C-Delta4Ms)can effectively improve the fault localization accuracy of MBFL and greatly reduce localization cost.In both single-fault and multiple-fault programs,C-Delta4 Ms improve accuracy by more than 79.5%compared with the MBFL technique,while the localization cost is less than45% of the MBFL technique.The Delta4 Ms technique and CTCR strategy alone can also significantly optimize the localization efficiency and accuracy,and perform better than MBFL technique.