Abstract
The recent popular game, Pokemon GO, created two symbiotic social networks by location-based mobile augmented reality (LMAR) technique. One is in the physical world among players, and another one is in the cyber world among players' avatars. To date, there is no study that has explored the formation of each social network and their symbiosis. In this paper, we carried out a data-driven research on the Pokemon GO game to solve this problem. We accordingly organised the collection of two real datasets. For the first dataset, we designed a questionnaire to collect players' individual behaviours in Pokemon GO, and used maps of Melbourne (Australia) to track and record their usual playing areas. Based on the data that we collected, we modelled the formation of the symbiotic social networks in both physical world (i.e. for players) and cyber world (i.e. for avatars) as well as interactions between players and Pokemon GO elements (i.e. 'bridges' of the two worlds). By investigating the mechanism of network formation, we revealed the relatively weak correlation between the formation processes of the two networks. We further incorporated the real-world pedestrian dataset collected by sensors across Melbourne CBD into the study of their symbiosis. Based on the second dataset, we examined the changes of people's social behaviours in terms of most visited places. The results suggested that the existence of the cyber social network has reciprocally changed the structure of the symbiotic physical social network. (C) 2017 Elsevier B.V. All rights reserved.