基于存储库数据挖掘的开源软件成功度量方法  

Open Source Software Success Measurement Method Based on Mining Software Repository

作　　者：宁德军 叶培根[2] 刘琴[1] 李梅[3] NING De-jun;YE Pei-gen;QIN Liu;MEI Li(School of Software Engineering,Tongji University,Shanghai 201210,China;School of Communication and Information Engineering,Shanghai University,Shanghai 201210,China;Shanghai Advanced Research Institute,Chinese Academy of Science,Shanghai 201210,China)

机构地区：[1]同济大学软件学院,上海201210 [2]上海大学通信与信息工程学院,上海201210 [3]中国科学院上海高等研究院智慧城市研究中心,上海201210

出　　处：《电子学报》2018年第12期2930-2935,共6页Acta Electronica Sinica

基　　金：中国科学院战略性先导科技专项(No.XDA06000000);国家重点研发计划项目(No.2016YFB1000805);上海市"科技创新行动计划"基础研究重点科技项目(No.17511107303;17511110202);上海市扬帆计划(No.17YF1420500)

摘　　要：开源软件已广泛应用于各软件领域,例如操作系统、容器等,但目前尚没有一种能够综合度量开源软件的方法.我们在用户兴趣度量和开发参与度量的基础上,提出了一种能够克服度量维度单一的局限性的度量方法.基于DM模型、软件生存力模型等相关文献研究和开源软件存储库数据挖掘,通过对项目过程数据进行聚类、主成分分析、回归分析和对开发过程的思考,本文提出一种基于存储库数据和统计学算法的开源软件成功度度量模型.并通过与用户兴趣度量结果和开发参与度量结果对比分析,证明本文的度量模型能够基于可自动无扰采集的存储库数据,更全面的衡量开源软件项目的成功.度量模型可应用于企业选择优质开源项目、学术研究、智能项目推荐等领域.The open source software(OSS)is widely used in various software fields,such as operating system,container,etc.But there is no method to measure open source software comprehensively.Based on the measurement of user interest and the development of participation metrics,we propose a method that can overcome the single limitation of the metric dimension.Based on DM model,software viability model and other relevant literature research,mining of software repository,we consider the OSS development process and propose an OSS success evaluation model through clustering,principal component analysis and regression analysis.By comparing the metric score of user interest method and development participation method,the evaluation model can be used to measure the success of OSS projects based on the data collected automatically without interference.The evaluation model can be applied to select high quality open source projects,academic research,intelligent project recommendation,etc.

关 键 词：开源软件 主成分分析 成功度量模型 数据挖掘 量化分析 软件工程 DM模型 软件生存力模型 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]