Multidimensional Characterization of Evolutionary Clusters

Software architects regularly have to identify unwanted couplings between different parts of a software system. Identifying groups of software entities which frequently changed together in the past, i.e. evolutionary clusters, are one way to help find such unwanted couplings. However, there may be many such evolutionary clusters. Not all of them point to unwanted couplings. In this chapter we discuss how a multi-dimensional characterization of evolutionary clusters can help identify unwanted couplings. In addressing this question we describe (1) properties used for characterizing evolutionary clusters, (2) scenarios characterizing unwanted couplings, and (3) the mapping of such scenarios to queries on a set of evolutionary clusters, resulting in a subset denoting unwanted couplings according to that scenario. We apply the proposed characterization to the case of a large embedded software system having a development history of more than a decade. By doing so, we execute Step 4 of the process described in Section 1.7 to help the software architect.