Deep neural network-based music user preference modeling, accurate recommendation, and IoT-enabled personalization

Abstract

With the popularity of personalized recommendation systems, how to better satisfy users’ emotional needs has become a key issue in the recommendation field, especially in the Internet of Things environment, where real-time access to users’ emotional data brings new challenges to recommendation systems. Existing recommendation methods primarily depend on users’ historical behavior or content-based features. However, they often overlook the impact of emotional states on recommendation effectiveness, which limits the adaptability and personalization of traditional systems. To solve this problem, this study proposes an emotional music recommendation system based on deep neural networks, which combines emotion modeling and hybrid recommendation strategies to provide more accurate recommendations. By combining user emotion data and music emotion features acquired by IoT devices in real time, our model can adjust the recommended content in real time, which significantly improves the emotion matching and recommendation accuracy. Experimental results demonstrate that the hybrid recommendation model significantly outperforms traditional content-based filtering (CBF) and collaborative filtering (CF) methods across multiple evaluation metrics, particularly in emotion matching (0.82) and recommendation accuracy (0.83). This study provides new ideas for emotion-driven personalized recommendation and technical support for future implementation of emotional recommendation systems in IoT environments.