CLUSTERING OF SOFTWARE MODULES WITH MULTIPLE VIEWS FOR EFFECTIVE MAINTENANCE COST

Abstract

Software clustering is the process of combining multiple systems or applications into a cluster that act as a single system. The subsystems are reverse engineered to extract design data. When large software systems are reverse engineered there is possibility of the system decomposition hierarchy. This hierarchy shows the system’s subsystems, the contents of the subsystems like modules or other subsystems. The Bunch’s software clustering tool shows how meta-heuristic search algorithms can be applied to the software clustering problem, successfully. But some uncertainty prevails whether Bunch provides optimal solution. The optimal solution for trivial systems may be achieved using an exhaustive search. Hence a clustering of software modules method with single graph is used in the existing method to obtain optimal solution by low coupling and high cohesion criterion. The advantage of using spectral methods is that the results this technique produced are within a known factor of the optimal solution. But the spectral single graph shows all the clusters in a single MDG which is complex and costly to maintain. The single graph view shows every system as a node which enables further complex cluster views. This study proposes a method called clustering of software modules with multiple graph views which improves the optimal efficiency and reduces the complexity. In this study the usage of multiple graphs for separate systems also reduces the maintenance cost to a greater extent.