Personalized Recommendations of Locally Interesting Venues to Tourists via Cross-Region Community Matching

You are in a new city. You are not familiar with the places and neighborhoods. You want to know all about the exciting sights, food outlets, and cultural venues that the locals frequent, in particular those that suit your personal interests. Even though there exist many mapping, local search, and travel assistance sites, they mostly provide popular and famous listings such as Statue of Liberty and Eiffel Tower, which are well-known places but may not suit your personal needs or interests. Therefore, there is a gap between what tourists want and what dominant tourism resources are providing. In this work, we seek to provide a solution to bridge this gap by exploiting the rich user-generated location contents in location-based social networks in order to offer tourists the most relevant and personalized local venue recommendations. In particular, we first propose a novel Bayesian approach to extract the social dimensions of people at different geographical regions to capture their latent local interests. We next mine the local interest communities in each geographical region. We then represent each local community using aggregated behaviors of community members. Finally, we correlate communities across different regions and generate venue recommendations to tourists via cross-region community matching. We have sampled a representative subset of check-ins from Foursquare and experimentally verified the effectiveness of our proposed approaches.