摘要
软件项目两个源文件之间的引用调用数量可视为源文件间依赖关系的强度,在软件开发的提交信息中两个源文件同时被修改提交的次数视为其同时修改的频度.我们证实有依赖的源文件更容易在同一次提交中被修改,即源文件的依赖关系强度和它们的同时修改频度具有一致性,并发现大量被同时修改而没有任何依赖的源文件往往存在问题,影响软件质量.本文用了复杂网络中社区划分的方法来研究软件工程中的问题,分别根据两源文件的依赖和同时修改关系,构建依赖关系网络和同时修改关系网络,使用社区发现算法(Girvan-Newman算法)对网络进行社区划分,对比分析两种网络对应的社区,并对不一致的现象探究原因,做出解释.
The number of references between two source files can be regarded as the strength of dependency of them,and the number of commits in which two source files are co-changed is considered as the co-changes between the two files. We found that dependent source files tend to be modified in the same commit,i. e.,the strength of the dependencies of the source files and their co-changes are consistent. Files which are co-changed frequently but without any dependencies often have problems that affect the quality of the whole project. This paper used the GN algorithm,which is used in community detection for complex networks,to study software engineering problems. Specifically,we used the number of references and the number of co-changes between two files as their edge weight to construct the dependency network and co-change network of software systems,respectively;then we used the GN algorithm to perform community division on the two networks;at last we analyzed the results of community division on the two networks and found out the inconsistence of the community division results of the two networks and explain it.
作者
胡学飞
李兵
李增扬
HU Xue-fei;LI Bing;LI Zeng-yang(School of Computer Science,Wuhan University,Wuhan 430072,China;Research Center of Complex Networks,Wuhan University,Wuhan 430072,China;School of Computer Science,Central China Normal University,Wuhan 430079,China)
出处
《小型微型计算机系统》
CSCD
北大核心
2019年第9期1981-1986,共6页
Journal of Chinese Computer Systems
基金
国家重点研发计划项目(2017YFB1400602)资助
国家自然科学基金项目(61572371,61702377)资助
武汉黄鹤英才现代服务业人才计划项目(230100007)资助
