Understanding Structural Hole Spanners in Location-Based Social Networks: A Data-Driven Study

Abstract

Location-based social networks (LBSNs) have been popular around the world. Some recent studies have focused on using online/offline social interactions among individuals to explain social phenomena, strongly demonstrating that data collected by LBSNs can be utilized to analyze the behavior of users. However, how structural hole spanners (SHS) behave in a LBSN requires more investigation. In this paper, we crawl the entire social network and all published tips of Foursquare, a leading LBSN app with more than 60 million users, using a distributed approach. Based on the crawled massive user data, we discuss the behavior characteristics of SHS in demographic, spatiotemporal and linguistic aspects. We further develop a classification model to accurately identify SHS and ordinary users based on their behavioral data. Our model achieved a high classification performance, with an F1-score of 0.821 and an AUC value of 0.879.