Design of Music Heritage Database System Based on Collaborative Filtering Algorithm

Abstract

In order to understand the design of music heritage database system, a research on the design of music heritage database system based on collaborative filtering algorithm is put forward. In this paper, firstly, collaborative filtering recommendation algorithm is studied, and users are recommended through two models that realize the algorithm, namely user-based collaborative filtering and item-based collaborative filtering. Finally, the recommendation effects of the two models are compared. Secondly, based on collaborative filtering, the content-based music recommendation algorithm is integrated, and a music recommendation system is designed. When the traditional collaborative filtering algorithm is used to recommend to users, some popular songs will affect the similarity calculation between users and between songs, resulting in poor recommendation effect. In this paper, the influence of popular songs on similarity calculation is reduced by punishing popular songs, and the accuracy of recommendation is improved by improving similarity calculation. Finally, the experimental results on the crawled Netease cloud music data show that the accuracy and recall of the hybrid recommendation algorithm are better than those based on collaborative filtering and content.