A couple of weeks ago, I have read this paper: "Friendship and Mobility: User Movement In Location-Based Social Networks". The paper is published in the proceedings of the 17th ACM SIGKDD Conference on Knowledge Discovery and Data Mining. The paper is mainly concerned with answering the following questions: (a) Do users' friendship affects their mobility ? (b) Do users' mobility make them create new friendship? (c) Do users visit specific places at particular times? Do they visit some places more frequently than others?
The Contribution of the paper is two-fold: (1) First, the authors have applied extensive data analysis on three data sets; two data sets from location-based social networking websites (i.e., Gowalla and BrightKite). The third data set is from a cell phone service provider in a European country the authors decided not to disclose its name (2) Secondly, the authors derived a user mobility model taking into account three sub-models: (a) Model of spatial locations that a user regularly visits, (b) A model of temporal movement between these locations, and (c) A model of movement that is influenced by the social network ties.
The authors figured out that distant friends have higher influence on a user than his nearby friends. This is explained by the fact that a user most probably travels (move long distance) to see friends. On the other hand, a user not necessarily visits a friend when s/he moves small distances (e.g., commute to work). Another finding was that users with highly similar trajectories are most probably to be friends.