As a special form of extracurricular sports activities, sports homework is an important part of school sports activities, which is an extension of the sports classroom, and can cultivate young people’s sports interests and sports habits in a subtle way. This paper proposes to build a personalized sports recommendation system to achieve online personalized guidance for secondary school sports homework. The system is mainly composed of three parts: user model, sports model, and recommendation algorithm. In order to improve the recommendation effect, by weighted integration of differential evolution algorithm and collaborative filtering algorithm, the system can recommend appropriate sports prescription according to students’ own characteristics and sports preferences. The recall, accuracy and coverage of the hybrid recommendation algorithm are 0.169, 0.036 and 0.547, respectively, which have obvious recommendation advantages compared with a single algorithm. The system in this paper relies on S Middle School to carry out experimental research, and after the experiment, there is a significant difference (p<0.05) between the students' sports quality and physical fitness under the guidance strategies recommended by the system, and the system in this paper ensures the accuracy of the recommendation results compared with manual guidance.
