Session-based Recommendation with Heterogeneous Graph Neural Networks
Lei Xu, Wu-Dong Xi, Chang‐Dong Wang · 2021
The aim of session-based recommendation is to predict the next-clicked item based on the anonymous behavior sequence. The existing works on session-based recommendation mainly capture the user preference within an individual session. This paper proposes a novel approach, called Session-based Recommendation with Heterogeneous Graph Neural Networks (SR-HGNN) to exploit cross-session information for better inferring the user preference of the current session. Specifically, we propose to use a heterogeneous graph to model the current session sequence and cross-session information simultaneously. After that, we come up with a novel model to pass messages along edges of different types hierarchically. Extensive experiments conducted on three real-world datasets demonstrate the superiority of SR-HGNN by comparing with different state-of-the-art baselines.