A Technique Based on Naming Patterns for Finding Candidates to Components from Source Code

Abstract

Component identification is relevant in maintenance activities but can be hard in scenarios where the volume of source code is high and there is no architecture documentation. This work presents a technique for finding concepts as candidates of architectural components. The technique is based in the frequency of terms that follows naming patterns that are proposed in this work. Projects from Eclipse and Apache, were used for evaluating the technique. The results show terms which match with some popular component types and domain concepts of the software in treatment. The technique has used source code without any architectural documentation showing that could be applied in emergent architectures.