Merging Inhomogeneous Proximity Sensor Systems for Social Network Analysis
Proximity information is a valuable source for social network analysis. Smartphone based sensors, like GPS, Bluetooth and ANT+, can be used to obtain proximity information between individuals within a group. However, in real-life scenarios, different people use different devices, featuring different sensor modalities. To draw the most complete picture of the spatial proximities between individuals, it is advantageous to merge data from an inhomogeneous system into one common representation. In this work we describe strategies how to merge data from Bluetooth sensors with data from ANT+ sensors. Interconnection between both systems is achieved using pre-knowledge about social rules and additional infrastructure. Proposed methods are applied to a data collection from 41 participants during an 8 day pilgrimage. Data from peer-to-peer sensors as well as GPS sensors is collected. The merging steps are evaluated by calculating state-of-the art features from social network analysis. Results indicate that the merging steps improve the completeness of the obtained network information while not altering the morphology of the network.