Recommendation of Massive English Teaching Video Resources Based on Knowledge Graph

Massive English teaching video resources contain a large amount of videos and user data. However, users only rated or interacted with a few videos. This leads to data sparsity, making it difficult to accurately infer the relationship between user interests and videos. In addition, how to select high-quality and suitable resources for students’ learning from numerous videos is also a challenge. In order to solve this problem, a recommendation method of massive English teaching video resources based on Knowledge graph is proposed. Using the Node2Vec method of network representation learning to determine the knowledge graph walking sequence nodes, introducing the neural network language Word2Vec method to achieve low dimensional mapping processing of sequence nodes, obtaining feature vectors of student users and learning materials, and establishing a knowledge graph. Naive Bayes classifier is used to classify the resources to be recommended. Determine the similarity of various English teaching videos using the similarity function, and then embed it into a matrix decomposition model to determine the feature matrix of student users and learning materials, achieving the recommendation of massive English teaching video resources. The experimental results show that the classification performance of Bayesian classifier is better, the average absolute error of this method is relatively stable, and students’ English learning performance is significantly improved with the application of the research method.