A Graded Software Defect Location Model Based On PageRank And Improved Genetic Tabu

The purpose of software testing is to ensure the quality of software,and software defect localization and test case generation are key research topics in software testing.Spectrum based software defect localization methods have certain effectiveness in locating software defects,but most existing automatic software defect localization methods have poor performance in solving multiple defect localization problems.Some effective methods still need further improvement due to their high complexity and the need for more interaction between developers.Moreover,existing defect localization methods mainly focus on the entire program,There is a problem of coarse granularity that prevents precise positioning.In response to such issues,this article focuses on the functional and statement level localization of software defects,and the main work is as follows.Firstly,a hierarchical software defect localization model framework is proposed to address the complex execution process caused by the entire program as the research object for defect localization.The defect localization process is divided into two parts: first locating the defect function,and then locating the defect statement.A hierarchical software defect localization model based on PageRank and improved genetic taboo mixing was designed.Secondly,in order to realize defect function location,this paper proposes a function level defect location algorithm based on improved PageRank.This algorithm first uses the function call graph generated when executing test cases to establish adjacency matrix,and then uses the improved PageRank algorithm to conduct preliminary function level location to reduce the complexity of defect location operations.Thirdly,in order to further accurately locate the defective statements on the basis of defect function set,this paper proposes a sentence level defect location algorithm based on improved genetic tabu mixture,and designs an adjustment method of dynamic population mutation rate in this algorithm to more accurately locate defective statements.Finally,the proposed model and algorithm were experimentally validated on the dataset Seimen,and comparative experiments and analysis of experimental results were conducted.