Research On Personalized Recommendation Method Based On Micro Ontology And Graph Matching

With the rapid development of information technology,information blocking is becoming more and more seriously,and the quality of users' demand for the information is also improved.These problems have become urgent problems to be solved.With the emergence of recommender system,academia and industry have a wide range of research and application in this field.But the current information recommendation,the results of information matching and similar content in various fields are not ideal compared with the information content that users need,and the resources that users really need can not be accurately recommended to users.Therefore,how to make users get reasonable and personalized resources is the main research goal of recommendation algorithm.This paper studies and analyzes these challenges,and proposes a personalized recommendation method based on micro ontology and graph matching,which improves the recommendation effect effectively.The main contributions of this paper are as follows:(1)According to the semantics of ontology technology,this paper first uses k-means method to cluster the domain information and user information,so that the micro ontology framework is composed of "concept,attribute and instance",and then the micro ontology framework can more clearly express the semantic relationship between the information,which can make the ontology structure clear and easily construct the micro ontology Body model.(2)According to the micro ontology model of "concept,attribute and instance",the topic micro ontology and user micro ontology are constructed into RDF data graph and query graph respectively.In the context of RDF graph,different paths represent the semantic relationship between vertices.In order to better match between micro ontologies,the path index pattern graph is introduced.(3)In view of the rationalization and personalization in the process of recommendation,the adjacent elements and semantic relations between the micro ontology are introduced,and a personalized recommendation method based on the matching of the micro ontology and graph is proposed.By establishing RDF path index of the adjacent elements with semantic relationship and their relationship for the subject ontology and user micro ontology,it is regarded as the matching between data graph G and query diagram Q when the subject micro ontology and user micro ontology match.First,the query graph q is decomposed into a group of paths,and a set of candidate matching paths are obtained for each path during the decomposition process;then the candidate paths are connected together by N-partition intersection graph.Finally,the approximate result set of the matching graph of the query graph is constructed,and the recommended results are obtained.Finally,the personalized recommendation method based on micro ontology and graph matching is carried out on movielens data set.The personalized recommendation method based on micro ontology and graph matching is compared with other algorithms.The experimental results show that the proposed method can improve the accuracy of the recommendation and is better than other algorithms.