Case Selection and Retrieval
Sankar Kumar Pal, Simon Shiu · 2004
Chapter 3 deals with the tasks of case selection and retrieval. It begins with the main issues of construction of similarity measures by defining first a few classical well known similarity measures in terms of distance, followed by the relevance of the concept of fuzzy similarity between cases and some ways of computing them. Methods of computing the feature weights, in this regard, using both classical, neural and genetic algorithm based approaches are then discussed. Finally, various methodologies of case selection and retrieval in neural, neuro-fuzzy and rough-neural frameworks are described. Here both layered network and self-organizing maps are considered for learning in supervised and unsupervised modes, and experimental results demonstrating the features are given.