Research And Application Of Personalized Recommendation System Based On Knowledge Graph

With the advent of the era of big data,while enjoying the convenience of the Internet,people are also facing the problem of information overload.Faced with the huge amount of data on the Internet,people cannot quickly and clearly distinguish between redundant data and modules of interest,and the recommendation system can meet the needs of users' personalization and customization,and can present the information that best meets the interests of users,so The recommendation system became the best solution.Traditional recommendation systems have sparseness problems and cold start problems,resulting in poor recommendation quality.The recommendation system can solve the problems of sparsity and cold start in the traditional recommendation system by referring to auxiliary information,such as context information,user or item attribute characteristics.Knowledge graph is also used as auxiliary information.Therefore,this article uses knowledge graph to enhance the recommendation performance by integrating knowledge graph with recommendation system.In this paper,a model(KG-RS model)that integrates knowledge graph and recommendation system is designed.The recommendation module and the knowledge graph embedding module are connected through the connection module,and the project and entity features are jointly learned in the connection module,which enhances the semantic information of the item and improves Performance of recommendation tasks;This article uses the Java language to design and implement the front-end and back-end of a personalized movie recommendation system page based on user functional requirements.The main research work of this article is as follows:(1)Analyze the research status of recommendation systems and knowledge graphs,and elaborate on the related technologies involved in this article,including the principles of common algorithms for recommendation systems,technical methods for knowledge graph construction,and similarity calculation of recommendation algorithms.The system evaluation index AUC,accuracy,precision and recall calculation methods are briefly introduced.(2)Aiming at the problem of sparseness of the recommendation system data and the recommendation accuracy is not high,this paper proposes a knowledge graph-based recommendation system model(KG-RS model).The model is divided into three parts.And connection modules.The KG-RS model uses the function of knowledge graph to embed low-dimensional vectors.The knowledge graph is embedded in the entity of the module and the items of the recommendation module are jointly learned in a lowdimensional space to establish the connection between the entity and the item,and then the items after the feature learning And entity are mapped back to the recommendation module and the knowledge graph embedding module,respectively.Finally,in the recommendation module and the knowledge graph embedding module,the user prediction value is calculated through a multilayer neural network.In this paper,the KGRS model is compared with other models on the MovieLens-1M data set.The experimental results show that the KG-RS model can improve the recommendation accuracy and perform best in other recommended evaluation indicators.(3)Based on the model proposed in this paper,a set of personalized movie recommendation system is designed and implemented.The system adopts a three-layer structure.Its front and back ends involve related technologies such as mybatis,spring,springmvc,easyui,etc.,and the system designs and implements movies.Website homepage,login registration,movie rating collection,movie search,personalized movie recommendation,website management six modules.