Similar Bug Identification Method Based On Bug Report And Source Code

In recent years,with the rapid development of software technology,the scale of software systems and development teams have also grown rapidly.Therefore,the number of bug reports to be repaired by developers has also multiplied,and it is difficult for developers to fully understand the entire code file of the whole software system.Therefore,improving the distribution efficiency of defect reports and locating the code files to be amended to improve the efficiency of repairing defects is a problem that needs to be solved urgently.Most of the current defect report and code file matching methods directly calculate the correlation degree based on the matching degree between the defect report and the code file,and sort all the code files based on the correlation,and recommend the top ranked code file for the defect report to be repaired.However,for the new bug report to be repaired,the similar bug report can be found from the historical bug report collection,and then the corresponding code file can be collected and finally the code file set similar to these code files can be found as a candidate correction code file for this new bug.Conversely,based on the current method of matching the bug report with the source code,it is possible to improve the practicability by using similar bug reports and similar code file identification methods.Specifically,the main contributions of this paper are as follows:First of all,this paper proposes a similar bug report identification method based on verb phrase and topic model.In addition to using commonly used vector space model methods to represent textual information,syntactic analysis,which is used to extract structured information of bug reports,is an important basis for measuring the similarity between reports.Secondly,using the similar defect report identification to improve the existing method for locating the code file to be repaired for the defect report.Compared with the existing research,this method considers the guidance of the historical defect report set to repair the current defect,and improves the present There is a method of recall that improves the efficiency of software repair.Finally,improve the vector feature construction method of existing similar code detection methods,consider the influence of program capacity on experimental results,improve the accuracy of existing methods,and use the hierarchical structure of abstract syntax tree and local sensitive hash algorithm to reduce the computation the complexity.In addition,the utility of the method in different high-level programming languages is also verified.