Research On Collaborative Filtering Recommendation Algorithm Based On Knowledge Graph Mate-path

With the rapid development of the Internet and information technology,the problem of information overload has seriously disrupted people's choice of information.Information overload means that the data on the network is beyond the range of people's acceptance and processing.To help people get the information they need quickly and accurately,recommendation systems have emerged.The recommendation algorithm is the core content of the recommendation system.The traditional recommendation algorithm calculates the similarity between users or items based on the user-item scoring matrix to implement the recommendation task.However,the traditional recommendation algorithm faces the following problems: First,due to the sparse data in the user-item scoring matrix,the traditional recommendation algorithm cannot accurately obtain similar users or items;Second,the traditional recommendation algorithm ignores the intrinsic attribute information of the item itself,so that it is not possible to find the specific interests of the user.Third,the traditional collaborative filtering algorithm can not explain the recommendation results.Fourth,the traditional collaborative filtering algorithm has the problem of cold start.In view of the above problems,this paper takes the recommendation algorithm as the research content.We introduce the concept of knowledge graph and meta-path into the traditional collaborative filtering algorithm,and proposes a collaborative filtering algorithm based on knowledge graph meta-path to improve the accuracy of recommendation results and explain the results of the recommendation to the user.The main work of this article is as follows:(1)This paper takes film recommendation as an example to construct a knowledge graph in the field of film.First,knowledge system of the film knowledge graph is constructed by analyzing the knowledge graph of film field.Then based on the movie data set Moive Lens-1M,the required knowledge is extracted from the open-link database DBpedia,and then the user's interactive behavior records and the extracted movies knowledge is integrated into the knowledge graph.Finally,We use the graph database Neo4 j to store the film knowledge graph.(2)In order to alleviate the problem of data sparsity faced by the traditional recommendation algorithm,this paper introduces two concepts of knowledge graph and meta path based on the traditional collaborative filtering algorithm,and proposes and implements a collaborative filtering algorithm based on knowledge graph meta-path(KG＿User CF).Specifically,KG＿User CF calculates the user's interest distribution matrix in all aspects of the film by mapping the meta-path in the film knowledge graph.Then calculate the user's interest in the film,and then generate a recommendation list for the user considering the user's various interests.Finally,the recommendation list generated by KG＿User CF is integrated with the recommendation list generated by the traditional collaborative filtering algorithm based on the user to generate the final recommendation list.At the end of this paper,through the experiments on the real world movie data set Moive Lens-1M,and compared with several algorithms,the effectiveness of the KG＿User CF algorithm proposed in this paper is verified.