Identifying key developers in software projects using code authorship metrics

Abstract

Code authorship is a key information in software projects. However, its practical usage in such projects is not widely explored. Therefore, in this thesis, we first define several authorship-centric concepts, which we use to investigate the development teams of 115 open source projects, including an in-depth analysis of the Linux kernel. After, we use code authorship metrics to address two well-known software engineering problems: (1) to assess knowledge concentration in software projects and (2) to identify skilled developers to maintain specific source code files. To address the first problem, we propose an algorithm to estimate truck factors (TF), a concept widely used by practitioners to reveal key project members. We use this algorithm to detect TF events in 1,932 projects and to reveal the practices that help them to overcome such events. Finally, to address the second problem, we investigate the effectiveness of authorship metrics to identify skilled maintainers in 10 projects.