Next Point-of-Interest Recommendation With Adaptive Graph Contrastive Learning
Xuan Rao, Renhe Jiang, Shuo Shang, Lisi Chen, Peng Han, Bin Yao, Panos Kalnis
Abstract
<italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Next point-of-interest (POI) recommendation</i> predicts user’s next movement and facilitates location-based applications such as destination suggestion and travel planning. State-of-the-art (SOTA) methods learn an adaptive graph from user trajectories and compute POI representations using graph neural networks (GNNs). However, a single graph cannot capture the <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">diverse dependencies</i> among the POIs (e.g., geographical proximity and transition frequency). To tackle this limitation, we propose the <underline xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><i>A</i></u><italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">daptive</i> <underline xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><i>G</i></u><italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">raph</i> <underline xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><i>C</i></u><italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">ontrastive</i> <underline xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><i>L</i></u><italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">earning</i> (<italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">AGCL</i>) framework. AGCL constructs multiple adaptive graphs, each modeling a kind of POI dependency and producing one POI representation; and the POI representations from different graphs are merged into a <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">multi-facet representation</i> that encodes comprehensive information. To train the POI representations, we tailor a <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">graph-based contrastive learning</i>, which encourages the representations of similar POIs to align and dissimilar POIs to differentiate. Moreover, to learn the sequential regularities of user trajectories, we design an attention mechanism to integrate spatial-temporal information into the POI representations. An explicit <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">spatial-temporal bias</i> is also employed to adjust the predictions for enhanced accuracy. We compare AGCL with 10 state-of-the-art baselines on 3 datasets. The results show that AGCL outperforms all baselines and achieves an improvement of 10.14% over the best performing baseline in average accuracy.