Personalized Recommendation Method of Sports Online Video Teaching Resources Based on Multiuser Characteristics

Abstract

Aiming at the problems of poor precision, low recall rate, and large recommendation time overhead in the personalized recommendation of sports online video teaching resources, this paper designs a personalized recommendation method for sports online video teaching resources based on multi-user characteristics. The area where the collected sports online video teaching resources are collected is fixed, and the confidence space for data collection is determined. The components of each clustering point are determined by the k-means clustering algorithm, and the data collection is completed continuously iteratively. The similar data in the data segments is removed with the help of cosine similarity, and the video segment data in the sports online video teaching resources with high similarity is removed for further data normalization. According to the multi-user feature analysis, the video teaching resource data that the user is interested in is matched, and the matching weight is calculated. In this paper, a personalized recommendation model of sports online video teaching resources is constructed by the recurrent convolutional neural networks (RCNN) method, and the personalized recommendation of video teaching resources is realized. The experimental results show that the mining accuracy of this method is about 96%, the highest recall rate can reach 98%, and its data recommendation time overhead is less than 1.3 s. This method effectively improves the effect of sports online video teaching resources.