Design and code reuse based on fuzzy classification of components

A bottleneck in software reuse is the classification schema and retrieval method of components. Particularly when large repositories of components are available, classification and retrieval for reuse should be flexible to allow the selection also of components which, although not perfectly matching requirements, are adaptable with a limited effort. This paper presents a fuzzy classification model for a repository storing descriptors of components. These descriptors include fuzzy-weighted keyword pairs describing components functionalities extracted from code and its design documentation. A mechanism for semi-automatic extraction of keywords, and for automatic assignement of fuzzy weights to keyword pairs based on text retrieval algorithms is provided.