Mining software repository to identify crosscutting concerns using combined techniques

Abstract

Modularization is a goal difficult to achieve in software development. Aspect mining aims to identify crosscutting concerns in non aspect oriented software allowing organizing them into aspects and, thus, improve the modularization. Several techniques have been proposed to identify crosscutting concerns from software repositories, usually by analyzing static data or execution traces. In this paper we present an approach on mining crosscutting concerns using combined results from two techniques: Frequent Closed Itemset Mining and Commit Frequency Analysis, which is presented in this paper as well. We evaluated our approach and observed that combining both techniques lead to better sets of crosscutting concerns candidates.