Software Clustering

This is an old revision of the document!

Software clustering researchers have developed several evaluation methods for software clustering algorithms. This research is important because

1. Most software clustering work is evaluated based on case studies. It is important that the evaluation technique is not subjective.
2. Evaluation helps discover the strengths and weaknesses of the various software clustering algorithms. This allows the development of better algorithms through addressing the discovered weaknesses.
3. Evaluation can help indicate the types of system that are suitable for a particular algorithm. For instance, Mitchell et al. in “CRAFT: A Framework for Evaluating Software Clustering Results in the Absence of Benchmark Decompositions” paper think that Bunch may not be suitable for event-driven systems.

The importance of evaluating software clustering algorithms was first stated in 1995 by Lakhotia and Gravely in “Toward experimental evaluation of subsystem classification recovery techniques” paper. Since then, many approaches to this problem have been published in the literature. These can be divided in two categories:

1. Based on an authoritative decomposition
2. Not based on an authoritative decomposition