Identifying Knowledge Divergence by Vocabulary Monitoring in Software Projects

During the development of a project, words used in source code add up to a big vocabulary, which may lead to a divergent word-understanding and word-knowledge between developers. Even the drop out of a single developer may lead to a big loss of knowledge about words and their meaning. By keeping track of the active developers vocabulary one is able to identify and react upon such situations, e.g., by applying pair programming to spread the knowledge around the team. In this work we propose a way to identify such situations by analysing the words contained in identifiers obtained through the commit history in a version control system. Initial empirical results are presented and analysed.