Improving design smell detection for adoption in industry

Abstract

This work identifies and analyses some barriers to improve the use of Design Smells detection in industry. First, the lack of agreement between tools and experts is established and analyzed for new information to help justify the differences in criteria used in identification. In this work we find indications that the information that describes the context of the analyzed project can help in this problem. Second, the excess of detected Design Smells is analyzed and addressed by proposing the prioritization of God Classes identified, as opposed to the all-or-nothing strategies of current tools, to help developers order its maintenance operations. This prioritization is validated according to this criterion.