Multiple Fault Location Based On Density Clustering

With the rapid development of computer technology,computer software is increasingly integrated into people's daily lives,and people are paying more and more attention to the security and stability of software.At the same time,the scale of software is enlarged and complicated,which makes the improvement of the quality of software code become a difficult point in software engineering.In the software testing process,developers often want to find out the latent error code and fix it in the software as soon as possible,thus reducing or avoiding the risk and economic loss caused by software defects.Software testing is an important means of improving the quality of software code.When a software defect is found,the development or tester needs to manually check the program code to find the specific location of the error code that caused the failure.This work requires a lot of manpower and time overhead.Therefore,how to quickly and accurately detect and locate error codes in software is widely recognized.In recent years,software fault location is a hot topic at home and abroad,with important scientific significance and application prospects.So far,the existing software fault location method is not perfect enough.The following problems exist:(1)The positioning effect of the existing fault location method is not ideal when there are multiple faults in the program,and it is difficult for the tester to locate the faults efficiently and accurately;(2)Existing multi faults localization methods mostly use partition-based clustering methods.However,the clustering method based on partitioning requires artificially setting the number of classes,that is the number of faults;(3)Less consideration is given to the effect of coincidental correct test cases on multiple faults location effects.Accidental correctness test cases can reduce the accuracy of fault location.In view of the above problems,this paper starts with density clustering and combines the identification of coincidental correct test cases for software fault location research.Firstly,for the multi faults program,a method based on density clustering is proposed,which performs adaptive clustering without preset the number of faults.Secondly,the influence of coincidental correct test cases on the fault location efficiency is studied.Analyze and propose the method of coincidental correct test case identification.Finally,it is studied how to combine the density clustering algorithm with the identification of coincidental correct test cases to optimize the clustering effect and improve the accuracy and efficiency of fault location.The main contributions of this paper are summarized as follows:(1)Propose an improved density clustering algorithm to cluster without preset faults number;(2)Accurately identify coincidental correct test cases in multiple faults procedures;(3)Optimize the density clustering algorithm to process the coincidental correct test cases to better cluster and improve the fault location accuracy.Case studies of 3240 fault versions of nine programs show that the proposed technique performs better in terms of fault location and clustering effects and efficiency than other cluster based multi faults location techniques.