Impacts of mobility and social networks on social activity-travel participation using location-based social network data

This paper investigates the relationship between social networks and the activities they generate, by exploring inter-social-activity durations as a proposed measure of social activity participation frequency. To model the proposed measure, data were collected and processed from a publicly-available dataset sourced from the location-based social networking service Gowalla. The data include information from 3065 Texas Gowalla users, regarding social activity-travel behavior, and performance of modularity- and surprise-based community detection. To account for the longitudinal nature of the data, and for possible spatial instability of the model parameters across two major Texas cities, a grouped-random-parameters hazard-based duration modeling approach with heterogeneity in means is employed, and separate models are estimated for Austin and Dallas users. The results suggest that social activity participation frequency is affected by individual mobility, and by a number of social network effects, such as ego social network size, social group variety, and local closeness centrality. The findings call for a thorough investigation of the transportation system and social network interrelationships.