Summarizing software concerns

While working on software concerns to perform evolution tasks, developers often encounter a lack of abstraction. They have to work with all the details of large subsets of code that constitute a concern at a low level of abstraction. In this paper we propose a framework to summarize software concerns in order to raise the level of abstraction and to subsequently improve the productivity of software developers. We use a combination of static analysis, information retrieval, and natural language processing techniques to extract and deduct knowledge about different parts of the concern and its interactions with other concerns. Then we produce a description of the concern using natural language generation and summarization techniques.