Research on Recommendation Methods for Scientific and Technological Information and Their Application in College Education - Based on Knowledge Graphs

In the current technology information recommendation system, the construction of the user behavior matrix frequently results in matrix sparsity and user cold start issues. To address this problem, TransAR-CF, a knowledge graph-based technology information recommendation method, is proposed, which mainly combines item similarity and semantic similarity to form a hybrid similarity matrix and then accomplishes the task of constructing a knowledge graph-based technology information recommendation model. The model is applied to a college education by constructing a college education service system, and the recommendation model and system experience are tested separately. The convergence speed of this paper’s model is higher than that of the CFKG model on both datasets A and B, and the difference between the two ranges from 0.02 to 0.12. According to the four dimensions of system design, learning support, evaluation design, and teaching effect, 70% to 90% of users have a positive attitude toward this paper’s system. In summary, the TransARCF method solves the problem of scientific and technological information that is difficult to adapt accurately but has high acceptance and good utility in college education.