Abstract: With the rapid development of smart education and digital teaching, how to quickly and accurately understand students' mastery of knowledge points has become a problem to be solved for intelligent teaching. The research uses the improved collaborative filtering algorithm to construct an intelligent recommendation evaluation model for test questions, realizes the intelligent recommendation of students' test questions and automatic evaluation of answering results, and conducts application experiments on the intelligent recommendation model. The experimental results show that the highest hit rate of the improved collaborative filtering algorithm is 89%, and the average hit rate is 15.6%. Compared with the traditional filtering algorithm, its hit rate is increased by 5.4%. In the practical application test, the model can quickly obtain the results and scores, and according to the students' scores and errors, it can provide students with targeted questions and learning suggestions. The recommendation effect of the improved collaborative filtering recommendation algorithm is higher than that of the traditional algorithm, which can effectively increase teachers' understanding of students' knowledge, reduce teachers' workload, and help teachers improve teaching quality.

Key words: intelligent teaching, collaborative filtering algorithm, intelligent recommendation, evaluation