Mining Location-based Social Networks: A Predictive Perspective

With the development of location-based social networks, an increasing amount of individual mobility data accumulate over time. The more mobility data are collected, the better we can understand the mobility patterns of users. At the same time, we know a great deal about online social relationships between users, providing new opportunities for mobility prediction. This paper introduces a noveltyseeking driven predictive framework for mining location-based social networks that embraces not only a bunch of Markov-based predictors but also a series of location recommendation algorithms. The core of this predictive framework is the cooperation mechanism between these two distinct models, determining the propensity of seeking novel and interesting locations.