A Music Recommendation System Based on Psychotherapy

Music can arise emotions and feelings easily, and thus has a strong influence on people's mental status. However, focusing only on the genre features of the music itself, traditional music recommendation systems often ignored the close relationship between music and emotion. Furthermore, the “filter bubble” phenomenon can easily make users' mental status even worse by recommending sorrowful music when they are depressed. In this study, we designed a novel music recommendation system based on psychotherapy. Specifically, our LSTM-based model can not only select the most helpful music based on users' previous mood and current emotion stimulus, but also use the care factor to adjust the results in order to improve users' mental status. The empirical experiments and user study demonstrated the effectiveness and usefulness of our proposed system.