Research On A Technique Of Multiple Fault Localization Based On Conditioned Classification Execution Slicing Spectrum

Software products are an indispensable part of people's lives.The reliability and usability of software products have become an important issue of concern.People put forward higher requirements for software testing and debugging in the ever-changing environment and needs.Debugging is one of the means to improve the quality of software,it can effectively find out the software faults.In recent years,the diversity and complexity of software have brought great challenges to software debuggers.Therefore,accurately identifying the faults in the software is the research hotspot of current software fault localization.The spectrum-based fault localization(SBFL)is one of the most popular software fault localization technologies.The technology locates the fault localization by collecting the coverage information and execution result of the program.However,the SBFL method does not take into account the dependency of the program,and there are limitations.A technique of conditioned execution slicing spectrum-based multiple fault localization(CESS-MFL)considers the dependency of the program and can reduce the efficiency of the program as the number of faults increases.However,this technique is still greatly affected by the fault-independent statement.Aiming at the above problems,this paper proposes a conditioned classification execution slicing spectrum-based multiple fault localization(CCESS-MFL).This technique improves and classifies the predicate conditions in CESS-MFL technology,and designs a set of suspiciousness formulas to calculate the suspicious value of program elements.The CCESS-MFL technology performs condition feature set according to predicate condition and fault correlation execution slicing,classifies conditional feature set to obtain conditioned classification execution slicing spectrum,then calculates the suspiciousness of the element,and finally generates suspiciousness report.CCESS-MFL technology uses the seven sub-kits of the standard dataset Siemens to verify its effectiveness.Experiments show that it is superior to the current popular Tarantula,Jaccard,Ochiai and CESS-MFL technologies,which can further reduce the influence of irrelevant statements associated with a fault on other fault statements.