Object-oriented software architecture recovery using a new hybrid clustering algorithm

In order to recover high-level software architecture from existing systems, we define Weighted Directed Class Graph(WDCG) to represent object-oriented software in this paper, which not only reflects static information of lowest level composition of software but also reflects dynamic information of software running. A new hybrid clustering algorithm based on hierarchical clustering and partition clustering is proposed for recovering high-level software architecture from WDCG. Four metrics are introduced to measure the effect of the new clustering algorithm for software architecture recovery. Experimental results show that our algorithm performs best in terms of software clustering quality, authoritativeness and extremity of cluster distribution.