Software Modules Clustering Using Social Network Algorithms

Abstract

Software development life cycle continues even after deployment, and changes or enhancements made after deployment are the most complex. Especially, in the absence of design documentation, implementing these changes can be tricky as it can adversely affect the design thereby disturbing software’s modular structure. So, it is imperative to quickly analyse and understand the modular structure of the software. Software Module Clustering Problem (SMCP) is one possible method to understand structure of complex software systems and enhance it without degrading the modular structure and violating the design rules. Various Artificial Intelligence(AI) techniques have been applied in the past providing an optimal modular structure, but they are time consuming. This research models SMCP as a community detection problem inspired by social networks. The results indicate that using community detection algorithm(CDA) an optimal solution can be achieved in terms of time and produces near optimal results for modularization quality.