基于贡献者行为特征的开源软件缺陷预测研究

Research on open source software defect prediction based on characteristic behavior of contributors

针对开源软件存在缺陷,改善软件质量等问题,文章提出了基于贡献者行为特征的开源软件缺陷预测研究。首先获取Apache软件基金会中的开源软件项目,运用Git和SVN版本控制系统对开发人员日志信息提取;然后采用K均值聚类算法模型挖掘开发人员团队(贡献者),采用词频统计和主成分分析算法模型得到贡献者行为特征;最后利用随机森林算法实现对贡献者特征行为的软件缺陷预测,该实验结果具有一定的参考意义。

Aiming at the problems of open source software defects and improving software quality,this paper puts forward a research on open source software defect prediction based on the behavior characteristics of contributors.First,obtain the open source software project in the Apache Software Foundation,and use GIT and SVN version control system to extract the developer’s log information;Then,the K-means clustering algorithm model is used to mine the developer team(contributors),and the word frequency statistics and principal component analysis algorithm model are used to obtain the behavior characteristics of contributors;Finally,the random forest algorithm is used to predict the software defects of contributors’characteristic behavior.The experimental results have certain reference significance.