PGST: Using Personal and Global Factors in the Spatio-Temporal Domain for Mobility Relationship Measurement

As well-developed social networking services collect rich location data of mobile users, it becomes interesting for measuring the mobility relationship strength. The state-of-the-art measure for such relationship is Personal, Global and Temporal factors (PGT) method, which calculates personal and global background in the spatial domain. We argue that meeting events differentiate from each other not only in the spatial domain, but also in the temporal domain. Accordingly, a novel method called PGST for measuring the mobility relationship is proposed. Additional time differences are introduced into the personal and global factors. Thus, background effects of location and time are removed from primitive meeting events. Experiments on real dataset suggest that our method significantly outperforms the state-of-the-art methods.