Mining Natural Language in Code Review Comments

Presenter: Vasiliki Efstathiou
Date: 15 June 2017


The growing availability of open software repositories has advanced research on mining software engineering data. Besides code-specic data, an interest towards developer communication data has emerged aiming to uncover features that impact on the software development lifecycle. Code reviews in particular, provide rich textual communicative information directly coupled with edits in source code. This talk will discuss possible unexplored directions towards analyzing natural language in code review comments, by adapting simple ideas from the community of linguistics. The proposed research aims to identify natural language patterns that imply higher level semantics, related to the underlying reasoning and intentions (such as necessity and probability), of the message conveyed in the comment. The ultimate goal in this context, is to uncover associations between the high-level semantics of comments and the revisions they suggest, and potential effects in comment usefulness.