Research on Intelligent Recommendation of Electronic Resources Based on Big Data and Knowledge Graphs

Abstract

At present, it is not easy to meet the needs of users with electronic intelligent recommendations. This paper proposes a research design aimed at conducting intelligent recommendation research on electronic resources. According to the knowledge graph theory, we will determine the overall process of constructing a knowledge graph of electronic resources based on the web crawler used to obtain the research data in this paper. Based on the integrated data, we will use the JAVA development language and Neo4j graph database to construct the knowledge graph of electronic resources, which will then be stored in a dataset format. By combining matrix decomposition and knowledge graphs, we create intelligent recommendations for electronic resources and conduct experiments on them. The model reduces the loss value score to 0.097 after 200 iterations. The precision rate of this paper’s model is 0.5614, the recall rate is 0.9540, and the value of the F1-score is 0.7068, which is significantly better than SVM (Support Vector Machines), DT (Decision Tree), RF (Random Forest), and GBDT (Gradient Boosting Decision Tree). This paper is capable of accurately recommending e-resources that are suitable for the user, and it perfectly realizes the concept of personalized intelligent e-resource recommendation.