Open Source Galaxy: Heterogeneous Information Networks in Social Coding

The GitHub has become the source of data for various research efforts. The full list of public events data which concern social coding on public repositories on this platform is available and can be analyzed. Modeling the complex nature of interactions among users on GitHub is challenge as well as dealing with such a huge amount of events data. Most existing research efforts on events data analysis are conducted on homogeneous networks without distinguishing different types of objects and links. In this paper, we design a data infrastructure to support the process of massive events data generated from GitHub and we model the complex interaction nature in collaborative developments as a heterogeneous information network, meta paths are introduced to capture various semantic information. We also propose a general algorithm framework with the ability to aggregate task related features and can generate heterogeneous network instances at arbitrary scale. Then we present the visualization of the open source communities by means of networks analysis.