Conclusion stability for natural language based mining of design discussions

Developer discussions range from in-person hallway chats to comment chains on bug reports. Being able identify that touch software design would be helpful in documentation and refactoring software. Design mining is the application of machine learning techniques correctly label a given discussion artifact, such as pull request, pertaining (or not) design. In this paper we demonstrate simple example how works. We then show conclusion stability poor different artifact types projects. two techniques—augmentation context specificity—that greatly improve cross-project relevance mining. Our new approach achieves AUC 0.88 within dataset classification 0.80 cross-dataset task.