Music Individualization Recommendation System Based on Big Data Analysis

Abstract

This study discovers a certain complementary relationship between different algorithms after conducting a comprehensive and in-depth analysis of proposal algorithms. This study proposes a big data music individualization proposal method based on big data analysis, which integrates user behaviour, behaviour context, user information, and music work information, based on traditional music proposal methods; improves the collaborative filtering proposal algorithm based on user behaviour; and calculates the semantic similarity between lyrics, as well as the song co-occurrence similarity based on the user's music download history. Because the lyrics represent the thoughts and feelings that the song wishes to convey to the listeners, the proposal module is completed, and the music proposal system is realized, by combining the two different similar information, using the improved algorithm and the Hadoop distributed framework. The music similarity and label similarity are combined to alleviate the problem of cold start and data sparseness, and a mixed similarity calculation formula is proposed to calculate the similarity between music. The accuracy similarity of the big data music proposal model proposed in this study is improved by about 20% through experimental comparison when compared with the collaborative filtering model and the hybrid model. It reflects the efficiency, scalability, and stability of the music proposal system as well as the ability to meet users' individual music needs.