文件信息增强的方法级软件缺陷定位  

Method-Level Software Bug Localization Enhanced with File Information

作　　者：薄莉莉 朱程 李斌[1] 孙小兵[1] BO Li-li;ZHU Cheng;LI Bin;SUN Xiao-bing(School of Information Engineering,Yangzhou University,Yangzhou,Jiangsu 225127,China;Key Laboratory of Safety-Critical Software(Nanjing University of Aeronautics and Astronautics),Ministry of Industry and Information Technology,Nanjing,Jiangsu 211106,China)

机构地区：[1]扬州大学信息工程学院,江苏扬州225127 [2]高安全系统的软件开发与验证技术工业和信息化部重点实验室,江苏南京211106

出　　处：《电子学报》2023年第3期613-621,共9页Acta Electronica Sinica

基　　金：国家自然科学基金(No.61872312,No.61972335,No.62002309);南京航空航天大学科研基地创新(理工类)项目(No.NJ2020022);扬州大学“高端人才支持计划”(No.2019);扬州大学畜牧学学科特区学科交叉课题支持项目(No.yzuxk202015);江苏省“六大人才高峰”高层次人才项目(No.RJFW-053);江苏省“333”工程中青年科学技术带头人项目。

摘　　要：软件开发与维护中会产生大量缺陷报告,根据缺陷报告准确定位到缺陷代码的位置是极具挑战性的.目前大多数工作在文件粒度定位缺陷,虽然少量工作定位在方法粒度,但定位准确度较低.本文提出一个文件信息增强的方法级软件缺陷定位技术FMBL(a File information enhanced Method-level Bug Localization technology),考虑方法与文件之间的从属关系以增强缺陷定位准确性.通过综合考虑代码与缺陷报告的词汇相似度、语义相似度和代码长度度量它们之间的相关性.在六个开源软件项目上开展实验以评估FMBL的有效性.结果表明,本文方法在六个项目上的平均Accuracy@10、MAP(Mean Average Precision)和MRR(Mean Reciprocal Rank)值分别达到0.436、0.223、0.296.与现有方法BugLocator、BLIA(Bug Localization using Integrated Analysis)、BugPecker相比,本文方法在MAP指标上分别提升153.1%、209.1%、22.8%.A large number of bug reports are generated during software development and maintenance.Finding the lo⁃cation of the bug code based on the bug report is a challenging task.Most of the bug localization works focus on file granu⁃larity.Even though few works locate bugs in method granularity,the accuracy is relatively low.In this paper,we proposed FMBL(a File information enhanced Method-level Bug Localization technology),a method-level software bug localization approach enhanced with file information.It considers the affiliation between methods and files to enhance the accuracy of bug localization.Lexical and semantic similarity between code and bug reports,together with code size are used to measure the correlation of methods and files.We conduct experiments on six open source software projects to evaluate the effective⁃ness of FMBL.The results show that,with FMBL,the average Accuracy@10,MAP(Mean Average Precision),and MRR(Mean Reciprocal Rank)on the six projects can reach 0.436,0.223 and 0.296,respectively.Compared with BugLocator,BLIA(Bug Localization using Integrated Analysis)and BugPecker,our approach improves by 153.1%,209.1%and 22.8%in MAP,respectively.

关 键 词：软件维护 缺陷定位 词汇相似度 语义相似度 文件信息 方法粒度 

分 类 号：TP311.5[自动化与计算机技术—计算机软件与理论]