Research On Case Retrieval Strategy Of RS-CBR System For Knowledge Management

Nowadays in the times of knowledge economy, knowledge has been the core means of production of the society. The tacit knowledge accounts of a large proportion of part in general knowledge and it is closer to the step for knowledge value realization. Case is a kind of tacit knowledge. Case-based reasoning system is a case of self-managed learning system and Case retrieval is one of its core technology. The research of case retrieval is limited to case matching calculation method. However, A complete study on case retrieval includes processes such as case representation, organization and matching algorithm. This paper focuses these processes and related theory.Firstly, this paper introduced the research background of case retrieval strategy based RS-CBR system of knowledge management, drawn the significance of the research, reviewed the research situation at home and abroad, analyzed the main problem, and described the purpose, the organizational structure and the main content of the research.Secondly, this paper introduce the theory of KM、CBR and RS, including the concept of knowledge、the development process of KM, the concept of CBR、its application integration、its process model and the organic fusion between CBR and KM, the concept of RS、its application integration and the organic fusion between CBR and RS in case retrieval system.Thirdly, the specific application technology of RS is described in detail in case retrieval system. Considering the problem of continuous and missing data in case retrieval, this paper discussed data discretization method based information entropy and the method of filling vacancy data based RS.After the pretreatment research of the data, high operated RS method was chosen to reduce attributes. Then complete RS weight configuration and fuzzy-AHP were combined to calculate the final weight of attributes.Finally, the case retrieval strategy was researched from whole process angle. That is to say, the case retrieval strategy was researched based the method of case representation, structure and case index. After summarying all kinds of knowledge representation methods, this paper provided the physical structure of case with object-oriented application framework. Then the index of case was set up. In the research of case matching calculation, different case retrieval methods were chosen according to the scale of case base. The Nearest Neighbor is used to retrieve case when the case base is small; otherwise, the secondary search algorithm based on fuzzy C-means clustering is better.