Research On Case Representation And Retrieval Based On Ontology

Case-Based Reasoning is a reasoning model that utilizes old experience to solve new problems. Different from rules-based reasoning, CBR represents the past experience as cases, and solves the bottleneck of knowledge acquisition based on rule-based reasoning system. It requires no domain knowledge in advance, but combining with the guidance of domain knowledge, it can effectively improve the performance of CBR systems .In the traditional CBR systems, the cases and domain knowledge have a variety of representation models,but most are based on non-formal knowledge representation, and are difficult to restructure and expand the definition and organization of the cases and share the domain knowledge.Ontology is the hotspot in AI in recent years, and has been successfully applied in many fields such as information integration, semantic retrieval and so on. It is the formal explanation of the shared concepts, semantic relations between concepts in domain, and is a collection composed of the concepts, the characteristics of the concepts, the relationships between concepts, instances and axioms. Because of making decidable description logic as knowledge representation and having formal modeling language, Ontology becomes a general knowledge representation model. The ontology is introduced into CBR that can solve the above problems effectively.This article introduces the basic concepts and principles of CBR, and then introduces the definition of ontology, ontology modeling language, description logic , ontology construction and so on. On the basis of the research of CBR and ontology, this paper analyzes their characteristics individually and presents a way of ontology-based case representation and the corresponding way of case retrieval. In the ontology-based case representation, the cases are defined into three parts--case structure, case and domain knowledge, and the research and explanation are carried out on how to unified the three parts into ontology. In the case retrieval, the cases are firstly filtered, and then the case similarity is calculated. During the similarity calculation, the article introduces ontology similarity computation to improve the precision ratio of the case retrieval.Finally, based on the research, the verified experiments have been done, the results are analyzed, and the feasibility and effectiveness of the methods in this article are verified with experiments.