Artificial Intelligence-Based Hybrid Recommendation Algorithm for Learning Resources

Abstract

To meet the development needs of Chinese learners’ personalized learning in the context of “Internet+” educational applications, this paper proposes to fuse the topic model (LDA)with the convolutional neural network mode (CNN). It helps us obtain the topic information and semantic information of different learning resources. Then we can obtain the corresponding learning resource feature vector. Secondly, we use the TF-IDF and K-nearest neighbor algorithms to vectorize the different persona information and potential interests of the learners. Then we introduce above-mentioned learning resource feature vectors, use the cosine similarity of the multidimensional vectors to calculate the learning interaction degree between the learners and the learning resources. Finally, we use a collaborative filtering recommendation algorithm based on the K-nearest neighbor algorithm to combine the knowledge point feature vectors of learning resources and learners’ results. This study can realize personalized recommendation of learning resources for learners.