Study On Methods Of Knowledge Search And Knowledge Recommendation Based On Knowledge Map

Knowledge map re-combinates people, resources and relationships in context as a network structure in order to create value. With the help of knowledge map, knowledge can be searched, cumulated and resued more easily in the process of knowledge management in organization. How to obtain the required knowledge accurately in massive enterprise knowledge database is a challenging task, so providing more effective knowledge query and knowledge recommendation tool is one of important technologies in enterprise knowledge management system. So far, retrieval system is mainly based on the keyword search and classification technology. The former cann't get a satisfied precision and recall for users, while the later requires users have a high awareness of knowledge structure and shows a bad effect to knowledge search and resue. In the mean time, most existing recommdation technologies used in the Internet community has a poor precision because of the lack of credible information of knowledge structure and user relationships in Internet community. Therefore, combine with knowledge map, this paper gives a series more effective knowledge search and recommendation methods.This paper researches methods of a knowledge search and knowledge recommendation based on Reconfigurable Ontology-based Knowledge Management System developed by our design team. First of all, this paper researches the problem of knowledge map modelling based on ontology which abstracts classes and instances as nodes in knowledge map and links the nodes in knowledge map using properties and other relatioships. Secondly, this paper researches means of imporving search results based on relationships between nodes in knowledge map. 1) utilize relationships between nodes in knowledge map to make query extension; 2) utilize relationships between nodes in knowledge map to extend related search results; 3) according to definition of ontology, give three relationship search algorithms which are relationships between class and class, class and instance and instance to instance. Thirdly, combine with chrarcteristics in social network constructed by knowledge source and knowledge network built by knolwdge in knowledge map, this paper improves the existing knowledge recommendation algorithm—Pearson algorithm. The new method improved the accuracy of Pearson algorthm. Finally, this paper gives the design and experiment according to those methods to verify their validity.