Efficient Fine-Grained Location Prediction Based on User Mobility Pattern in LBSNs

Location-Based Social Networks (LBSNs) have built bridges between virtual space and real-world mobility in recent years. The massive check-in data generated in LBSNs has made it possible to predict users' future check-in location, which has proved meaningful for e-commerce developments. Existing studies mainly focus on predicting the next check-in location with a coarse granularity, which shows limited performance in practical scenarios. In this paper, we propose a comprehensive approach based on user mobility pattern for predicting users' future check-in location at any fine-grained time in LBSNs. Firstly, user mobility patterns involving time periodicity, global popularity and user preference are analyzed. Then, a set of predictive features are extracted. Finally, the features are combined into a supervised scoring model and a classification model respectively in order to predict users' future check-in location. Extensive experiments on three real-world datasets demonstrate the efficiency and superiority of the proposed approach in various metrics.