A software component selection technique based on fuzzy clustering

The software component selection in component-based software development remains a challenging area for those who try to retrieve the optimal components from the existing component repository. The clustering techniques offer a solution for component selection problems but with major shortcomings like a prior specification of number of clusters, specifying the radius of clusters, finding appropriate representation for the feature vectors, choosing the correct distance metric, overlapping clusters and use of subjective judgment of application developers. This paper present a software component selection technique based on fuzzy clustering using the concept of fuzzy relations. The main contribution of the algorithm is that the components which are similar based on multi attributes are kept together, helping the decision maker to choose the right component set. The algorithm is validated with a help of a small case study.