Integration of case-based reasoning and fuzzy approaches for real-time applications in dynamic environments: current status and future directions

Abstract

This survey reviews recent researches conducted on the application of fuzzy approaches to Case-Based Reasoning (CBR) dealing with real-time applications. Fuzzy approaches have been effectively applied for knowledge representation, feature selection, and learning in CBR. Dealing with imprecise and uncertain knowledge, generalization, mining, and learning also in combination with low computational complexity are the main advantages of fuzzy approaches used in the CBR context. This paper presents and summarizes new findings on the integration of fuzzy approaches with CBR. The survey results highlight the advantages of fuzzy approaches in CBR for real-time applications. They show the current state of fuzzy-based CBR approaches. In addition, fuzzy approaches which are more operative for each operation in CBR are addressed. Those operations most contributing to the advantages of the fuzzy approach will be pointed out and detailed. Low accuracy, storage and computational challenges with a large amount of experiences and uncertainties are important issues in case of real-time applications. This paper proposes a general fuzzy-based CBR approach for real-time applications to benefit the advantages of previous approaches. Finally, some considerations of latest developments in fuzzy approaches which may be introduced as potential research directions for real-time applications are stated.