PhD Dissertation Proposal by DOAN Thanh Nam

Learning Latent Characteristics of Locations

using Location-based Social Network

The popularity of smartphones and wearable devices in recent years has helped to create new location based social networking (LBSN) sites for users to publish their visits (or check-ins) to different venues. The ease of posting and sharing visited locations with friends has attracted many users using these applications lately. The check-in feature does not only exist on LBSN sites, many major online social networking sites also adopt this feature to enrich their social interaction. For instance, Facebook Places allows users to update their visited locations in their timelines. Similarly, Twitter’s users could associate their tweets with geo-locations.

Although there are many research works studying check-in behaviors of users in LBSN sites, they mainly focus on spatial homophily, social homophily and distance effect. This proposal will elaborate two new effects in LBSN sites named neighborhood competition and area attraction. Through the proposal, we will show the existence of these two effects and introduce multiple probabilistic models to apply their characteristics to understand the latent properties of venues in LBSN. Moreover, using these effects can help us to increase the accuracy of several applications in LBSN.

About the Speaker