Social-DBSCAN: A Presence Analytics Approach for Mobile Users’ Social Clustering

Abstract

The ability to discover social groups that can be attributed to the learning activities taking place at target locations, provides an invaluable opportunity to understand the presence and movement of people within a university campus. Utilising density-based clustering of WLAN traces, we illustrate how granular social groups of mobile users can be detected within such an environment. The proposed density-based clustering algorithm, which we name Social-DBSCAN, has real potential to support human mobility studies such as the optimisation of strategies of space usage. It can automatically discover the duration of the academic term, the classes, and the attendance data. For the evaluation of our proposed method, we utilised a large Eduroam log of an academic site. We chose, as a proof concept, selected locations with known capacity. We successfully detected the regular learning activities that took place at those locations and provided accurate estimates about the attendance levels over the academic term period.