Software component capture using graph clustering

We describe a simple, fast computing and easy to implement method for finding relatively good clusterings of software systems. Our method relies on the ability to compute the strength of an edge in a graph by applying a straightforward metric defined in terms of the neighborhoods of its end vertices. The metric is used to identify the weak edges of the graph, which are momentarily deleted to break it into several components. We study the quality metric MQ introduced in [1] and exhibit mathematical properties that make it a good measure for clustering quality. Letting the threshold weakness of edges vary defines a path, i.e. a sequence of clusterings in the solution space (of all possible clustering of the graph). This path is described in terms of a curve linking MQ to the weakness of the edges in the graph. We describe a simple, fast computing and easy to implement method for finding relatively good clusterings of software systems. Our method relies on the ability to compute the strength of an edge in a graph by applying a straightforward metric defined in terms of the neighborhoods of its end vertices. The metric is used to identify the weak edges of the graph, which are momentarily deleted to break it into several components. We study the quality metric MQ introduced in [1] and exhibit mathematical properties that make it a good measure for clustering quality. Letting the threshold weakness of edges vary defines a path, i.e. a sequence of clusterings in the solution space (of all possible clustering of the graph). This path is described in terms of a curve linking MQ to the weakness of the edges in the graph. This work is supported by grants from Coopération FrancoQuébécoise (project U10-6) and NSERC (Canada).