Modern software are pieces of complex work delivered collaboratively by groups of software developers. These developers and their relationships formed by co-editing the same source code file constituting a developer social network. Cer- tainly, different projects have different shapes of developer social network which can be reflected by their statistical properties. An important observation is that the structure of the developer social network would influence the efficiency and quality of the development process because of different patterns and efficiency of the communication among developers. Therefore, it is necessary to deepen our understanding of developer social networks in the bigger social network analysis context.

My supervisor and I implement a system that integrates developer social network construction, visualization, analysis and recommendation functionalities proposed in this dissertation. This system helps researchers and project managers to identify existing problems with current developer social networks and provide suggestions by recommending appropriate relationships between developers to optimize the network. The system is implemented using Java and other third party utilities. Developer social network is a particular type of social network, which has its own properties such as its dynamics and its bipartite-graph nature. Our research explores some previously unexplored topics which will certainly deepen our understanding of developer social networks and provide a useful tool for developer social network analysis and optimization.