At present, the selection of teaching video by users is blind. To change this problem, an accurate recommendation method based on KNN collaborative filtering technology is proposed. We study the characteristics of online education resource courses, collaborative filtering algorithm and personalized recommendation technology. Aiming at the defects of traditional collaborative filtering technology, SVM algorithm with KNN algorithm are adopted for filtering and sorting, according to user preferences and needs, o overcome the impact of data quality on recommendation algorithms. Then, the weighted sum of the similarity of the item scoring matrix is also used to calculate the user similarity, and to push the similar teaching resources that users were most interested in. The results of case analysis showed that the collaborative filtering algorithm designed by us can make implicit evaluation on the system to modify the parameters of the system, effectively improve the accuracy of personalized recommendation and adapt to multiple changing user groups.