Graph hierarchical dwell-time attention network for session-based recommendation

Session-based recommendation (SBR) is making item recommendations based on anonymous click behavior. SBR graph neural networks has shown great power in recent years. It can enhance the representation of items a session. The aggregation then used to generate session vector for recommendation. However, existing models rarely consider impact dwell-time data when performing aggregation. contains implicit behavior users sequence. In order obtain more accurate embedding and take into account multiple perspectives, we propose new model, hierarchical attention network. This approach uses modified network learn by extracting loss information from modeling. We also design module that effect long-term preferences sessions. Experimental results show GHDAN outperforms state-of-the-art session-based methods.