A method based on naming similarity to identify reuse opportunities

Abstract

We propose a method for extraction of reuse opportunities based on naming similarity of two types of object-oriented entities: classes and methods. Our method, called JReuse, computes a similarity function to identify similarly named classes and methods from a set of software systems from a domain. These classes and methods compose a repository with reuse opportunities. We also present a prototype tool to support the extraction by applying our method. We evaluate the JReuse with 71 software systems mined from GitHub, in 4 domains distinct. In total, we evaluate 1,772,269 lines of code; 75,447 methods, and 17,821 classes. As a result, we observe that the JReuse is able to identify main classes and methods that are relevant in domain selected. Furthermore, we evaluated the JReuse, through a survey with developers GitHub, of each evaluated domain. The observed results indicate that our method is able of extracting the possibilities for reuse, independent of the software domain.