A Context and Emotion Aware System for Personalized Music Recommendation

Music recommendation systems are an emerging application that helps users to find their favorite music in numerous archives. Most existing music recommendation methods focus on exploring users' profiles, listening histories and audio signal of music to recommend the most relevant items to users. However, users' preferences may vary in different contexts or in response to changing emotions. In recent years, some studies have affirmed the important roles of context and emotions in music recommendation, and include context or emotions to their system design; however, few studies take both context and emotions simultaneously into consideration. In this paper, we propose an integrated approach to enhance the prediction of a user's preference; this approach incorporates the factors of context and emotion and aims to provide users with a more simple, intuitive and enjoyable listening experience. In addition, we adopt serviced-oriented architecture to implement our music recommendation system to which new innovative services can be easily added or integrated to provide more flexible services in the future. We also present the evaluation results of the prediction accuracy and users satisfaction.