Research And Application Of Movie Recommendation Algorithm Based On Knowledge Graph

With the development of internet technology,it is difficult for people to quickly find movies that interest them when facing a large number of films.Movie recommendation algorithms are the most effective means to help users quickly find movies that interest them.This article focuses on the problem of poor recommendation effect of collaborative filtering recommendation algorithm when facing data sparsity,and the problem of mainly relying on user historical behavior data for recommendation while ignoring item information.Firstly,a movie domain knowledge graph was constructed.Then,a knowledge graph representation learning method was introduced into the traditional collaborative filtering recommendation algorithm,and a knowledge graph-based collaborative filtering recommendation algorithm was proposed.Finally,the feasibility and effectiveness of the algorithm were verified through experiments,and a movie recommendation system was designed and developed based on practical application scenarios.The main research work and results include:1.A movie domain knowledge graph was constructed.Based on Movielens dataset,we use a crawler to crawl the TMDB movie website to obtain the data for building the knowledge graph of movies,use Protégé software to construct an ontology library,and form a triad of data for knowledge extraction of TMDB movie data and finally import it into Neo4 j graph database for storage and display.The experimental results show that knowledge mapping can make the data more intuitive and easy to understand,and it is also very helpful to discover the inner patterns of the data.2.Propose a collaborative filtering recommendation algorithm based on knowledge graph.The algorithm introduces the knowledge graph representation learning method into the traditional item-based collaborative filtering recommendation algorithm to obtain vector representations of entities and relationships,calculate the similarity between movies.At the same time,in order to solve the problem that the random negative sampling of Trans R representation learning method may sample low-quality negative triplets and affect the model training effect,this paper uses the K-Medoids clustering algorithm to divide the entities of triplets into different clusters,and selects the same cluster for replacement when performing negative sampling to reduce the generation of low-quality negative triplets and improve the quality of model training.By combining the similarity between items calculated based on the semantic information of knowledge graph items and the similarity between movies calculated based on user historical rating data,the algorithm generates predicted ratings and provides recommendations for users.The experimental results on the Movielens dataset show that the algorithm can effectively improve the recommendation efficiency.3.A movie recommendation system is designed and developed using Spring Boot+My Batis-Plus+Vue technology.The system includes user login and registration,personal information management,data collection and update,movie management and user recommendation,etc.The collaborative filtering recommendation algorithm based on knowledge graph proposed in this paper is successfully used for actual movie recommendation.The test shows that the system is fully functional and the user experience is good.