Intelligent Recommendation Method of Personalized Educational Resources Based on K-means Clustering

The information overload and other problems faced by learners are becoming more and more serious. Under the contradiction between the dynamic and personalized demand of learners for resources and the static teaching resources of the system, only the short-term learning interest of users is obtained, which leads to high sparsity and low accuracy after recommending resources. Therefore, a personalized educational resources intelligent recommendation method based on K-means clustering is proposed by this paper. Organize the knowledge points of personalized educational resources, including binding relationship checking, normalization and decomposition strategies, to ensure that the knowledge points are at the atomic level, which is convenient for learners to learn. The short-term and long-term learning interests of users are extracted by Embedding layer and K-means clustering algorithm, and the personalized learning interests of users are accurately extracted. The matching relationship between user interests and educational resources attributes is established, and the user's interest towards resource attributes is estimated by collaborative filtering and inverse frequency to solve the sparsity problem between knowledge points and user interests and educational resources. A model was built on the basis of the K-means clustering algorithm. The intelligent recommendation of personalized educational resources is realized by calculating the similarity between user behavior data and educational resource attributes. The comparative experiment results indicate that the method proposed improving the scores of students in G1 and G2 groups, and the improvement of scores is more obvious. At the same time, on the sparsity index, the sparsity calculated by the proposed method decreased by about 25% and 30% as a whole, which effectively solved the sparsity problem and improved the recommendation accuracy.