Enhancing Software Evolution through Design Pattern Detection

Software system evolutions can be supported through different techniques and by exploiting different tools. We concentrate our attention on the advantages we gain through design recovery, and in particular on sub-component recovery, which helps to detect logical components of the system and their relationships. Components can be of various kinds: an important category is that of design patterns. Several approaches have been proposed to automate design pattern detection. In this paper we describe our approach to design pattern detection using supervised classification and data mining techniques based on sub-components, and summarize the results we obtained on behavioral design patterns.