With the growing number of large scale software projects, software development and maintenance demands the close developer interactions. Having a thorough understanding of the group of developers is critical for improving software quality and reducing cost. In contrast to most commercial software endeavors, developers in open source software (OSS) projects enjoy more freedom to organize and contribute to a project in their own working style. Their interactions through various means in the project generate a latent developer social network (DSN). Examining the structure and evolution of these social networks as well as their similarities and differences from other more general social networks (GSNs) is of value to our software engineering community. It allows us to begin building an unde...[ Read more ]

With the growing number of large scale software projects, software development and maintenance demands the close developer interactions. Having a thorough understanding of the group of developers is critical for improving software quality and reducing cost. In contrast to most commercial software endeavors, developers in open source software (OSS) projects enjoy more freedom to organize and contribute to a project in their own working style. Their interactions through various means in the project generate a latent developer social network (DSN). Examining the structure and evolution of these social networks as well as their similarities and differences from other more general social networks (GSNs) is of value to our software engineering community. It allows us to begin building an understanding of how well the findings from other fields based on GSNs could be applied to DSN. In this thesis, we compare DSNs with popular GSNs such as Facebook, Twitter, Cyworld (a large social network in South Korea), and the Amazon recommendation network. We found, for instance, that while most social networks exhibit power law degree distributions, DSNs do not. We also investigate the similarity and differences among the DSNs extracted using different social linkage indicators (e.g., co-occurrence in bug reports). The findings facilitate a better understanding of DSNs and provide useful references for researchers who study DSNs. In addition, we also examine how DSNs evolve over time, highlighting how events within a project (such as a release of new software or the departure of prominent developers) impact the structure of the DSNs, and observe the evolution of topological properties such as modularity and the paths of communities within these networks.