A Software Defect Location System Based on Source Code File Dependency

Abstract

The present invention implements a software defect location system based on source code file dependencies by using methods in the field of artificial intelligence. The system is divided into three modules: input, calculation, and output. The input module is used to import defect reports and source code files, and the output module It is used to sort the source code files according to the correlation score and output them externally. The calculation module adopts the DependLoc framework and consists of three sub-modules. The CNN4TFIDF model sub-module captures text features according to the defect report and the TF-IDF vector of the source code file; the segment RefHI encoder The sub-module encodes defect reports and source code files into vectors with source code dependency characteristics; the CNN4RefHI sub-module is based on the correlation score between defect reports and RefHI vectors of source code files; thus effectively distinguishing error and non-error source code files; Covers all source code files in the current application; the technical effect of making full use of information in defect reports and TF‑IDF vector representations of source code files.

Description

Technical field [0001] The present invention relates to the field of artificial intelligence, and more particularly to a software defect positioning system based on source code file dependencies. Background technique [0002] Open source software typically records defects using defect tracking systems such as Bugzilla and Jira, and has a large number of defect reports per day. The defect report contains the description of the defect, the relevant program status, logs, etc. when the failure is expired. Therefore, the researchers attempt to automatically position the wrong program entity based on the submitted defect report. Defect positioning based on defect reports can be seen as a query problem, that is, for a given defect report (query), you need to find a possible error in all source files (documentation) of the application, and the suspicious source file file Rank according to the possibility of an error. In recent years, research work around the defect report is mainly divid...

AI Technical Summary

Problems solved by technology

[0009] The present invention aims at the defect automatic location problem based on the defect report, and aims to propose a defect location method based on source code file dependencies, so as to solve the problem that the existing method location is not accurate enough and file dependencies are not considered

[0010] Specifically, the main problems to be solved include: (1) Directly using the dependencies between source codes for positioning will introduce many irrelevant files, so it is necessary to find a quantification method for dependencies, which needs to meet two conditions: Can effectively distinguish error and non-error source code files; can cover all source code files in the current application

Images