Statistical Analysis of Social Coding in GitHub Hypernetwork

Li Kuang, Feng Wang, Heng Zhang, Yuanxiang Li

Abstract: Social coding is a software development approach, which can facilitate hundreds of developers collaborating in one project simultaneously. Many researchers focus on the analysis of social network based on complex network models. However, the traditional complex network model cannot express the full information of collaboration. In this paper, in order to depict the properties of hypernetwork well and get a good simulation result, we investigate the time evolution of the GitHub dataset. We find that, (1) The hypernetworks show high level of self-organization; (2) From the neighbor connectivity of developers, some of the skilled developers wish to collaborate with skilled developers, whereas some skilled developers prefer to collaborate with freshman; (3) From the statistical properties of programming languages communities, the assortativity of Java community is obviously different from other communities, and the projects have a high probability of collaboration with those using the same programming languages.