Dispersive prism

Image by Kelvinsong via Wikimedia Commons, CC0 (own work)

Speaker: Angelika Schmid

The project, that Angelika presented, was born in an interdisciplinary cooperation of the chair of Software Engineering of the IM department and the chair of Statistics from the WIWI department.

Version control data, such as GitHub data, can provide interesting insights into the dynamics of software developer cooperation and problems therein. With the ever growing computational power of modern systems, the possibilities to analyze large and complex networks grow as well. However, it is still challenging to gather any information from complex networks that goes beyond the estimation of parameters of previously known growth patterns. Spectral methods often take a successive snapshot perspective, instead of a integrated process perspective.

Spectral methods allow the handling of large networks and provide an important link between dynamic and topological descriptions of network growth. Depending on the way nodes and edges are conceived, and how relationships are defined, the methods allow us to trace the development of developer groups and thematic modules. The current results however indicate that the standard implementation of the method. This is not surprising, as developer cooperations are highly volatile and the spectral network growth model the research builds upon uses static snapshots instead of a process-oriented perspective.

During the talk, Angelika was providing insights into the research groups’ current efforts to adapt the method to a dynamic, process-oriented context, involving information on mails, commits, and other version-control related information.