Using data from location based social networks for urban activity clustering

Understanding the spatial and temporal aspects of activities in urban regions is one of the key challenges for the emerging fields of urban computing and emergency management as it provides indispensable insights on the quality of services in urban environments and helps to describe the socio-dynamics of urban districts. This work presents a novel approach to obtain this highly valuable knowledge. We hereby propose a segmentation of a city into clusters based on activity profiles using data from a Location Based Social Network (LBSN). In our approach, a segment is represented by different locations sharing the same temporal distribution of check-ins. We reveal how to describe the topic of the determined segments by modelling the difference to the overall temporal distribution of check-ins of the region. Furthermore, a technique from multidimensional scaling is adopted to compute a classification of all segments and visualize the results. The proposed method was successf ully applied to Foursquare data recorded from May to October 2012 in the region of Cologne (Germany) and returns clear patterns separating areas known for different activities like nightlife or daily work. Finally, we discuss different aspects related to the use of data from LBSNs.