Characterize geo-social properties of an online social network

This article tries to characterize the geo-social properties of an online social network from Foursquare within an integrated framework with domain knowledge from sociology and geography. A hierarchical structure of subunits, including individual, pair and group, is identified and the geo-social properties of each subunit are studied respectively. As the novelty of this article, the characteristics of group-level units in sociology and geography, which are named as community and cluster, are clarified. Two data mining techniques in community detection and spatial clustering are employed to identify the groups and the relationship between these two concepts is revealed based on clustering comparison. The power-law distribution of friendship possibility against distance is observed in this study, confirming previous researches. And the small spatial clusters in a social community are recognized both by visualization approach and clustering comparison parameters.