PhD Dissertation Defense by DOAN Thanh Nam | Learning latent characteristics of locations using location-based social networking data

About The Talk

With many users signing up location-based social networking (LBSN) platforms to share their daily activities, these platforms become a gold mine for researchers to study human visitation behavior and location characteristics. Modeling such visitation behavior and location characteristics can benefit many useful applications such as urban planning and location-aware recommender systems. In this dissertation, we focus on modeling two latent characteristics of locations, namely area attraction and neighborhood competition effects using LBSN data. Our literature survey reveals that previous researchers did not pay enough attention to area attraction and neighborhood competition effects. Area attraction refers to the ability of an area with multiple venues to collectively attract check-ins from users, while neighborhood competition represents the need for a venue to compete with its neighbors in the same area for getting check-ins from users. We firstly gather the LBSN data generated by Foursquare users from two big cities: Singapore and Jakarta. To generalize our findings, we also employ the Gowalla data of users from New York City. We then embark on a data science study of area attraction, neighborhood competition, and other user and location related effects including spatial homophily, social homophily, distance effects. Since the interaction between users and locations is a complex process involving multiple effects, we propose several novel models that incorporate latent location and social factors in the generation of users' visitation. These models utilize a range of different techniques, including PageRank, Bayesian reasoning, matrix factorization, and neural networks. Each model is evaluated through extensive experiments and the results show that neighborhood competition and area attraction effects contribute to more accurate modeling and prediction of users' visitation to locations.

Speaker Biography

DOAN Thanh Nam is a PhD candidate in School of Information Systems, Singapore Management University, working with Professor Ee-Peng Lim. He received his Bachelor of Engineering (Computer Science) from the Ho Chi Minh City University of Technology (HCMUT) in 2011. His current research focuses on using location-based social network to model the latent characteristic of venues to understand the check-in behavior of users.