An Emotion-oriented Music Recommendation Algorithm Fusing Rating and Trust

• Main Authors: Jiwei Qin, Qinghua Zheng, Feng Tian, Deli Zheng
• Format: Article
• Language: English
• Published: Atlantis Press 2014-04-01
• Series: International Journal of Computational Intelligence Systems
• Subjects: Similarity weight; Emotion; Trust; Music recommendation
• Online Access: https://www.atlantis-press.com/article/25868479.pdf

With the overwhelming increase of music, it has become difficult to find music which suits the taste of a listener who is in a certain state of emotion. Focusing on the listener's emotional state, this paper presents an emotion-oriented music recommendation algorithm. First, the listener's similarity is calculated by the rating value and the trust value. More specifically, based on the number of ratings, two thresholds are set to extend the calculation strategy of listener similarity weight to selectively use the trust value and correlation to the rating value. Second, because the music and the listener's emotional state are different objects, there is no obvious way to match one with the other. The listener's preference for emotional connotation of music is introduced to bridge the gap between the music and the listener, and that solves the issue of how to match the listener's emotion with music. Lastly, considering the difference of listener's perception of musical content and the complexity of the listener's emotional response, we propose a comprehensive measure to evaluate the accuracy, the coverage and the listener's satisfaction degree with the recommendation. Experimental results show that the presented algorithm comparing the collaborative filtering and trust-based recommendation, results in a tiny loss of accuracy with the improvement of larger coverage, thus not only obtaining a perfect tradeoff between accuracy and coverage, but also increasing the degree of listener satisfaction.