Optimizing retrieval process and using neural networks for adaptation process in case based reasoning systems

Abstract

The retrieval process in case based reasoning systems (CBR) is a two-step process. It starts with a problem description and ends when a best matching previous case(s) has/have been retrieved. To optimize the retrieval process, enhancement of both processes is required. This research work explores the use of XML (Extensible Markup Language) as a case descriptive language. An additional goal is to identify factors which play a major role in the optimization process. This work also presents an experimental investigation concerning the use of artificial neural networks in the adaptation process of CBR systems. A backpropagation feedforward neural network in different configurations, has been employed to carry out empirical analysis of using this technique for case based adaptation.