Optimization and Performance Evaluation of Personalized Algorithm in Music Recommendation System

Music recommendation system (MRS) is a system that recommends music content for users based on their historical behavior and interests. Personalized algorithm is the core technology of MRS, which provides personalized music recommendation for users by analyzing their behaviors and interests. However, there are some problems in the current personalized algorithm, such as too much concentration of recommendation results and lack of diversity of recommendation results. Therefore, it is of great significance to optimize and evaluate the performance of personalized algorithms. In this paper, experimental method and comparative method are adopted to analyze content-based, user-based and hybrid recommendation algorithms, and attention mechanism and multi-task learning in deep learning are taken as improvement methods to improve the accuracy and effectiveness of personalized recommendation. Experimental results show that the hit rate of the improved algorithm is up to 0.79. The improved method presented in this paper achieves good performance on multiple data sets.