The Weighted Combined Algorithm: A Linkage Algorithm for Software Clustering

Software systems need to evolve as businessrequirements, technology and environment change. Assoftware is modified to accommodate the requiredchanges, its structure deteriorates. There is increaseddeviation from the actual design and architecture. Veryoften, documentation is not updated to reflect thesechanges thus making it more and more difficult tounderstand, manage and maintain these systems.Researchers have applied various techniques to recoverthe components and architecture of such softwaresystems. The use of clustering techniques has recentlybeen explored for reverse engineering and softwarearchitecture recovery. There is a need to tailorclustering algorithms and similarity measures to caterto software. In this paper, we present a new algorithmfor finding inter-cluster distance. We compare theperformance of some popular similarity measures forthis algorithm using two test systems and suggestvariations of the similarity measures which show betterresults for software clustering.