Providing location recommendations has become an essential feature for location-based social networks (LBSNs), as it helps the users to explore new places and makes LBSNs more prevalent to them. Existing studies mostly focus on introducing the new features that affect users' check-in behaviours in LBSNs. However, despite the difference in the type of the features exploited, they mostly follow the same principle - characterizing dependencies between the probability of a user visiting a point-of-interest (POI) and each feature separately. The decision of a user on where to go in an LBSN, however, is driven by multiple features that act simultaneously. On the other hand, applying a full model which considers all the features jointly suffers from overfitting, as for each user there is limited available data. In this paper, we propose an intermediate solution by fragmenting the model into multiple partial models which each takes the subset of the features as the input. The proposed approach focuses on building the personalized partial models (PRMs) which are further combined by applying an additive approach. Experiments on two datasets from Foursquare show that our proposed method outperforms the state-of-the-art approaches in POI recommendation.
|Title of host publication
|2018 IEEE Global Communications Conference (GLOBECOM)
|Published - 21 Feb 2019
|IEEE Conference Proceedings