Software Architecture Module-view Recovery Using Cluster Ensembles

Abstract

Software architecture documents are valuable assets supporting the maintenance process for software systems. Unfortunately, in many projects, software architecture documentation is not conducted properly or the documents become obsolete due to a discrepancy with the current architecture. To address this, various automated methods to recover software architecture have been proposed in the literature. We argue that most previous studies have not considered cluster ensembles but relied on a single clustering algorithm. In this paper, we propose to take advantage of cluster ensembles for software architecture recovery. Our experiments on five open-source projects are reported and the results are analyzed.