Research On Recommendation Methods Based On Movie Knowledge Graph

In today’s streaming websites,movie recommendations are the main way for websites to guide users to watch movies.The significance of recommendation systems for movie websites lies in improving user satisfaction and loyalty,thereby increasing website traffic and revenue.The current movie recommendation methods applied in movie websites are mainly based on collaborative filtering to generate recommendation lists for users.However,research on enhancing the recommendation effectiveness by constructing a knowledge graph based on the content attributes of movie resources is relatively scarce.A knowledge graph can model and represent the relationships between movies,classify and describe movies from multiple perspectives,and provide more effective structural and content information for recommendation methods.This article focuses on the problem of movie recommendation based on knowledge graphs.It uses knowledge graphs,Trans E graph embedding,and knowledge graph convolutional techniques to mine potential connections between users and movies from knowledge graph data,enhance the interpretability of recommendations,and improve accuracy to better meet users’ needs and increase user stickiness.To better represent the potential connections between users and movies in the knowledge graph,this article adds hierarchical aggregation weight factors to the traditional recommendation model(Ripple Net)and proves through experiments that different propagation attribute layers in knowledge graph propagation have different influences on information aggregation.Based on the user-item bipartite graph,a collaborative filtering method based on the principle of structural traversal is used to find interest points that users may be interested in sparse data,enriching the representation of similar user information points.Secondly,based on the research on the hierarchical propagation information of the knowledge graph,the principle of graph convolutional neural network is used to propose an information representation method based on knowledge graph convolutional network(Irr KGCN).We use the knowledge graph convolutional network to collect preference propagation based on attention mechanism for information in the knowledge graph.We use the knowledge graph to generate user and movie profiles and finally calculate the cosine similarity between the two in the output layer of the model as the probability of user-movie interaction.Finally,we verify the effectiveness of the proposed method on the knowledge graph dataset constructed using Movielens.