Design And Implement Of Collaborative Filtering Recommendation System Based On Knowledge Graph Feature Learning

With the advent of the age of big data,the volume of information on the Web has increased in an exponential fashion,drowning users in the flood of countless products,news and movies,etc.Recommendation system,based on the historical interactive data of users,provides personalized recommendation services for users,which has become an indispensable and important tool to solve the problem of information overload.However,the traditional collaborative filtering recommendation algorithm only relies on the unstructured data of users,which has the problem of data sparsity.To solve this problem,many recommendation algorithms used structured knowledge graph data as auxiliary information to generate more accurate recommendation results.This thesis mainly describes the technical principles involved in the collaborative filtering recommendation system based on knowledge graph embedding and focuses on the discussion of the implicit feedback recommendation algorithm assisted by knowledge graph.Aiming at the shortcomings of semantic feature extraction of knowledge graph,the traditional knowledge graph embedding algorithm is improved and optimized,and finally the collaborative filtering recommendation system integrating knowledge graph is realized.The main work is as follows:(1)This thesis makes a comprehensive and systematic literature review on the technology research related to recommendation system in recent years,especially the research on knowledge graph assisted recommendation.The implicit feedback data modeling and knowledge graph embedding techniques involved in the collaborative filtering recommendation based on knowledge graph are summarized,and the related algorithms involved are analyzed and compared.(2)Aiming to the two key problems of structured knowledge acquisition and item semantic feature extraction and the sparse problem faced by collaborative filtering algorithm,this thesis proposed a collaborative filtering recommendation algorithm based on knowledge graph embedding.Firstly,knowledge information related to items was extracted from the Freebase Knowledge Graph and linked with historical interactive items to construct knowledge subgraphs.Then the Xavier-Trans R method based on Trans R obtained the representation of entities and relations in the sub-graph.An end-to-end joint learning model was designed to embed structured information and historical preference information into a unified vector space.Finally,the collaborative filtering method was used to further calculate these vectors to generate an accurate recommendation list.(3)Experimental results on two public datasets,Movielens-1M and Amazon-book,show that the proposed algorithm is superior to the baseline algorithm in terms of accuracy,recall,F1 value and NDCG metrics.It means that the above method can integrate large-scale structured and unstructured data,while obtaining high precision recommendation results.(4)A collaborative filtering recommendation system integrating knowledge graph was designed and implemented,in which Neo4 j graph database was used to build knowledge subgraph and Python was used to build system interactive interface.