社交编程网站的项目网络分析

Project network analysis of social coding site

社交编程网站如Git Hub,使得开发人员在互不认识的情况下,能够合作开发某个项目。这种软件合作开发方式的转变,使得对社交编程网络的研究很有价值。现有的研究很少有利用社区结构分析社交编程网络。通过Git Hub网站提供的数据构建了项目网络,结合重叠社区发现的算法,对该网络进行研究。研究发现:a)项目间的合作多数是基于小团体的模式;b)社区重叠部分的项目有较高的影响力;c)大规模社区的编程语言的组成和整个网络语言组成存在差异。

Social coding sites such as GitHub, enable developers to conduct software development cooperatively even though they don＇ t know each other. This changing of software development makes the study of social coding network extremely valuable. Previous studies rarely analyzed the social coding network with community structure. This paper constructed a project-project relationship network with data collected from GitHub and investgated the structure of the network by combining overlapping community discovery algorithm. It shows that：the cooperation of the project are mainly based on groupuscule model. Projects crossing multiple communities tend to be more influential. The kind of programming language and their distribution are different between the network and large-scale communities in it.