一种基于文本分类和评分机制的软件缺陷分配方法  

Software Bug Triaging Method Based on Text Classification and Developer Rating

作　　者：史小婉 马于涛[1] SHI Xiao-wan;MA Yu-tao(School of Computer Science,Wuhan University,Wuhan 430072,China)

机构地区：[1]武汉大学计算机学院,武汉430072

出　　处：《计算机科学》2018年第11期193-198,219,共7页Computer Science

基　　金：国家重点基础研究发展计划(973)(2014CB340404);国家自然科学基金(61672387;61702378);武汉市黄鹤英才(现代服务)计划资助

摘　　要：开源软件项目的缺陷管理和修复是保障软件质量及软件开发效率的重要手段,而提高软件缺陷分配的效率是其中亟需解决的一个关键问题。文中提出了一种基于文本分类和评分机制的开发者预测方法,其核心思想是综合考虑基于机器学习的文本分类和基于软件缺陷从属特征的评分机制来构建预测模型。针对大型开源软件项目Eclipse和Mozilla的十万级已修复软件缺陷的实验表明,在"十折"增量验证模式下,所提方法的最好平均准确率分别达到了78.39%和64.94%,比基准方法(机器学习分类+再分配图)的最高平均准确率分别提升了17.34%和10.82%,从而验证了其有效性。Bug management and repair in open-source software(OSS)projects are meaningful ways to ensure the quality of software and the efficiency of software development,and improving the efficiency of bug triaging is an urgent problem to be resolved.A prediction method based on text classification and developer rating was proposed in this paper.The core idea of building the prediction model is to consider both text classification based on machine learning and rating mechanism based on the source of bugs.According to the experiment on hundreds of thousands of bugs in the Eclipse and Mozilla projects,in the ten-fold incremental verification mode,the best average accuracies of the proposed method reach 78.39%and 64.94%,respectively.Moreover,its accuracies are increased by 17.34%and 10.82%,respectively,compared with the highest average accuracies of the baseline method(machine learning classification+tos-sing graphs).Therefore,the results indicate the effectiveness of the proposed method.

关 键 词：缺陷分配 文本分类 评分 预测模型 支持向量机 

分 类 号：TP311.5[自动化与计算机技术—计算机软件与理论]