Personalized music recommendation based on knowledge graphs and social circles

With the rapid development of digital music ser vices, personalized music recommendation systems have become crucial for enhancing user experience. Addressing the limitations of traditional recommendation algorithms in understanding user needs and recommendation accuracy, this study proposes a knowledge-based social circle personalized music recommendation model (KSPMR). This model aims to provide more comprehensive and accurate music recommendations by deeply analyzing individual user preferences. Initially, we developed a data preprocessing method to optimize embedding computation and graph construction efficiency. Subsequently, we explored the impact of social circles on user music preferences, integrating social network factors into the recommendation logic. Finally through a series of comparative experiments, the superiority of our model in improving recommendation effectiveness was demonstrated. Experimental results show that, compared to baseline models, our model significantly improves across various performance metrics, especially in accuracy and effectiveness of recommendations after incorporating information from the user and artist social circles.