Mining Communication Patterns in Software Development: A GitHub Analysis.

Background: Studies related to human factors in software engineering are providing insightful information on the emotional state of contributors and the impact this has on the code. The open source software development paradigm involves different roles, and previous studies about emotions in software development have not taken into account what different roles might play when people express their feelings. Aim: We present an analysis of issues and commits on five GitHub projects distinguishing contributors between users and developers, and between one-commit and multi-commit developers. Method: We analyzed more than 650K comments from 130K issues of 64K contributors. We calculated emotions (love, joy, anger, sadness) and politeness of the comments related to the issues of the considered projects and introduced the definition of contributor fan-in and fan-out. Results: Results show that users and developers communicate differently as well as multi-commit developers and one-commit developers do. Conclusions: We provide empirical evidence that one-commit developers are more active and more polite in posting comments. Multi-commit developers are less active in posting comments, and while commenting, they are less polite than when commented.