Leveraging Software Engineering Data: multiple sources, multiple opportunities

Prof Marcelo Maia (Universidade Federal de Uberlândia) will give a talk on Thursday 17 November 2016 11:00 AM in room A21 (Inria Lille Building A).

Title: Leveraging Software Engineering Data: multiple sources, multiple opportunities

Abstract: A rapidly increasing rate of new information being produced is a phenomenon observed in many different areas, such as, social interaction, scientific data, news. In Software Engineering the situation is not different. Software repositories evolved to deliver a social service. Version control systems were enhanced with new operations to allow more manageable forms of interaction. Issues are tracked during the life cycle recording lot of knowledge in a semi-structured way. The social interaction of developers all around the world is tracked in blogs, forums and question-and-answer (Q&A) sites generating partially structured content that can be considered one of the thriving forms of documentation available nowadays.
In this scenario, we face many challenges to explore the available opportunities. We have been studying techniques to assist developers in different facets of their work. Execution trace analysis could help to understand better your architecture, specially in poorly documented systems. Co-change analysis could help to improve modular quality. Longitudinal studies on change history could also be analyzed under the view of evolution of code anomalies. Source code available in repositories could be learned to predict repetitive tasks and assist API usage. Q&A is a source of raw knowledge that have applications having from redocumenting APIs to improve automatic repair techniques.