From proximity sensing to spatio-temporal social graphs

Abstract

Understanding the social dynamics of a group of people can give new insights into social behavior. Physical proximity between individuals results from the interactions between them. Hence, measuring physical proximity is an important step towards a better understanding of social behavior. We discuss a novel approach to sense proximity from within the social dynamics. Our primary objective is to construct a spatio-temporal social graph from noisy proximity data. We address the technical and algorithmic challenges of measuring proximity reliably and accurately. Simulations and real world experiments demonstrate the feasibility and scalability of our approach. Our algorithms doubles the sensitivity of proximity detections at the cost of a slight reduction in specificity.

Scalable Detection of Crowd Motion Patterns

“…While GPS receivers or video cameras could be used for this purpose, we explore another option. For this work, we follow the ideas by Martella et al and utilize proximity graphs to capture the texture of a crowd [3]. A proximity graph is a spatio-temporal graph where nodes represent proximity sensors and edges are proximity detections.…”

A Hierarchical Approach for Associating Body-Worn Sensors to Video Regions in Crowded Mingling Scenarios

“…As stated in Section 3, each of the wearable devices outputs a dynamic binary proximity graph, which is later refined to eliminate false neighbor detections using the method proposed by Martella et al [19]. Thus, for each time sample which is recorded at the same sample rate as the video (20Hz against 20fps) a proximity graph is created between the participants.…”

“…To refined false neighbor detections, they apply a density-based clustering to group all the neighbor detections in time, by comparing the graphs of consecutive times. This method leverages the bursty nature of the proximity graphs, meaning that the correct neighbor detections tend to appear sequentially together in time and the false detections tend to be isolated (see [19] for more details).…”

“…person leaving the field of view of the cameras and then returning). In addition, all experiments were done with a limited number of participants (19). All the above resulted in a rather limited evaluation.…”

Graph mining indoor tracking data for social interaction analysis

“…1). We can either infer the social interactions directly through the use of mobile proximity sensors; this person-person proximity setup produces a log of physical distance between tagged individuals from which a dynamic social network can be generated [8], [7]. The other option is the location-person WSN setup where an additional set of reference sensors are used.…”