Mining GitHub For Multiple Types Of Social Ties Among OSS Projects

As the internet technology has a strong effect on interchanging information between people around the world, new types of cooperation have arisen. Of these new forms, Open Source Software development has emerged where many geographically distributed developers work for different Open Source Software(OSS) projects. For sharing, archiving and exchanging information between developers, several software repositories are used. Currently, social coding sites, such as GitHub, are provided to gather these projects and developers in the same virtual platform. As a result, the collaboration between OSS projects becomes in the highest degree causing different social ties among them. This increases the importance of mining software repositories to detect these different relationships and study their features. But, the big challenging here is how we can extract useful information from these repositories and employ it in software engineering tasks in order to organize the collaboration among OSS projects.This thesis investigates networking OSS projects on GitHub as a large social coding site identifying three kinds of ties among these projects that hosted on it. For that, we collect the data of 108718 projects and 499485 users. In addition, we download the source code of 421 java projects that supported by Apache organization and hosted on GitHub. Accordingly, we construct three different project social networks(PSNs) in terms of the identified social ties and study their features using social network analysis(SNA) methods. Also, we make a comparison between the three constructed PSNs on two levels. i.e. Network Metrics level and Network Structure level in order to determine the similarities and differences among them.Results of this thesis help to comprehend different forms of collaboration among OSS projects and how they are actually correlated to each other. This constitutes the basis for developing systems that facilitate software engineering tasks like recommendation developers and for studying the aspects of coupling evolution to detect the impacted factors. Further, our study is a step toward formalizing laws that can organize the cooperation in open source software projects.