Abstract
Collaboration is important to software development processes and collaboration networks help us understand its structure and patterns. A common problem, however, is that these networks are not known and need to be discovered. In this work we study collaboration networks of five projects using an existing method that mines these networks from version control systems. The method is based on Recommender System techniques and finds similar developers by analyzing commits that are made to common files. These similarities are then used to automatically construct the network and it is visualized using a force directed graph layout algorithm. Two of the studied projects come from industry and are closed source while the other three are open source. In each study we learn some of the project's collaboration form and organization. We also were able to find various aspects of these projects that were previously not known. © Springer-Verlag Berlin Heidelberg 2013.