Research On Personalized Collaborative Recommendation System For Knowledge Resources Based On Ontology

As development of information technology and intensifying market competition, the demand of knowledge management for corporate organizations and individuals user is growing very fast, people urgently need information which could satisfy their knowledge demand from massive accumulation of knowledge resources of the original information system. Therefore, the personalized recommendation system of knowledge resources is increasingly becoming a hot research spot both in domestic and aboard. But the study of personalized recommendation systems for knowledge resources is relatively rare.In this paper, the principle and working processes of traditional personalized collaborative recommendation system which most widely used has been researched. In order to solve the problems of traditional user recommendation systems, a new personalized collaborative recommendation system of knowledge resource has been put up on the basis of Ontology. This paper firstly define and design a specific user interest model based on domain ontology, and then design the calculation way of the user interest rate and update mechanism of the user interest model, the mapping between customized personalized knowledge needs of users and the concepts which are from a special domain Ontology in user interest model. User interest model in this paper not only can be a good representation of the relationship of concepts in domain ontology, but also can demonstrate the user's individual needs of knowledge. The user interest model would be work as important part in the whole system. In the method of personalized collaborative recommendation system, we firstly take advantage of relations among concepts which belong to domain Ontology to predict the unrated knowledge resources instance through that were rated in order to solve the sparse problem of the user-instances matrix. On the basis of that, the existing collaborative recommendation algorithm to calculate similarity between two users can be more efficient, so the accuracy of the recommendation system could be work better than the traditional ones. Finally, the traditional user collaborative filtering recommendation system and the proposed collaborative personalized recommendation system were compared to proved by data verify that the improvements of precision of proposed system. the realization of prototype also can prove the feasibility of the proposed system .