Research Of Multiple Fault Localization Based On Cluster Analysis Of Program Failures

Software testing is absolutely necessary in the process of software development and maintenance,also is an important method for locating faults.With the development of the design complexity and diversification on software structure,locating faults is a very resource-consuming task,including the time and the cost.Therefore,mining associations-based fault localization technique becomes an important research direction.This mining association model considers the control dependence and the data dependence between statements,through the mining of failing execution relevant statements,to locate program fault.However,existing studies are based on a false assumption,ignoring that failing execution caused by different faults would affect the suspiciousness of the same statement.In actual process of locating faults,programs contain more than one faults and the number of faults is hard to predict.Hence,this paper is focus on how to divide the fault's space,and improving existing model to improve the efficiency and accuracy through converting the problem of multi-fault localization to the problem of single fault localization.Aiming at these problems,cluster analysis techniques of data mining techniques were deeply analyzed.And this paper introduces cluster analysis method of program failures,which is able to locate multiple faults in a program.The contents of this paper mainly include the following aspects:(1)A presentation of a new technique for cluster-analysis-based multiple fault localization(CA-MFL),which provides a new approach for categorizing the failures caused by different faults.The clustering analysis method helps in locating two or more faults in a program by illuminating likely faulty statements which belong to the same kind of failures.(2)We describe a target association algorithm and a corresponding way to review codes,according to the failure classes' differences,which are not restricted to a single way and can maintain the effectiveness of fault localization.(3)We systematically evaluate the effectiveness of cluster-based multiple fault localization on the SIR benchmarks under the same setting as previous studies.Four existing fault localization techniques were compared with our technique in this study,which demonstrates the superior accuracy achieved by our technique in fault localization.And the time and space complexity of our technique was analyzed.