Personalized Keyword Boosting for Venue Suggestion Based on Multiple LBSNs

Personalized venue suggestion plays a crucial role in satisfying the users needs on location-based social networks (LBSNs). In this study, we present a probabilistic generative model to map user tags to venue taste keywords. We study four approaches to address the data sparsity problem with the aid of such mapping: one model to boost venue taste keywords and three alternative models to predict user tags. Furthermore, we calculate different scores from multiple LBSNs and show how to incorporate new information from the mapping into a venue suggestion approach. The computed scores are then integrated adopting learning to rank techniques. The experimental results on two TREC collections demonstrate that our approach beats state-of-the-art strategies.