Next POI Recommendation based on Multi-graph Information Fusion and Attention Network

Abstract:
Next point of interest (POI) recommendation aims to model the user's check-in sequence and help users find the POI they are interested at the next moment. Recurrent neural network (RNN)-based methods have been widely used in next POI recommendation, but these RNN-based methods lack in-depth model of complex semantic relationships and deeper sequence transition relationships. We propose next POI recommendation based on multi-graph information fusion and attention network, named MGINAN. Experimental results on two real datasets demonstrate that our proposed method has better recommendation performance than baseline methods.