Selection of audio materials in college music education courses based on hybrid recommendation algorithm and big data

Abstract

Aiming at the problem of audio material selection in college music education courses and to solve the problem of students’ low enthusiasm caused by improper selection of audio materials, this paper adopts the hybrid recommendation algorithm combining big data, personalized recommendation algorithm based on Collaborative Filtering Recommendation Algorithm (CF). Big data is used to obtain user evaluation matrix, and Pearson correlation coefficient is used to calculate the similarity between users, so as to form the nearest neighbor set, obtain the nearest neighbor set of target users, and generate the user-based recommendation set. At the same time, a questionnaire is set up to obtain the real evaluation scores of each user on the audio materials. 20% of the questionnaire data are used for model testing and 80% for model training. The accuracy of the square root error is measured, and the prediction score obtained by the model is compared with the real score. It is found that the mean RMSE value of the model adopted in this paper is 0.3813, which is at least 2.564% higher than that of similar models, and has a higher accuracy. Meanwhile, the algorithm is relatively simple, providing a reference for audio selection of college students in courses.