Multi-View Enhanced Graph Attention Network for Session-Based Music Recommendation

Traditional music recommender systems are mainly based on users’ interactions, which limit their performance. Particularly, various kinds of content information, such as metadata and description can be used to improve recommendation. However, it remains addressed how fully incorporate the rich auxiliary/side information effectively deal with heterogeneity in it. In this paper, we propose a M ulti-view E nhanced G raph A ttention N etwork (named MEGAN ) for session-based learn informative representations (embeddings) pieces users from heterogeneous graph neural network attention mechanism. Specifically, proposed approach firstly models listening behaviors textual Heterogeneous Music Graph (HMG). Then, devised Attention Network is low-dimensional embedding by integrating enhanced multi-view HMG an adaptive unified way. Finally, hybrid preferences learned that satisfy real-time requirements recommended. Comprehensive experiments conducted two real-world datasets, results show achieves better performance than baselines, including several state-of-the-art recommendation methods.