Graph mining indoor tracking data for social interaction analysis

Abstract

With the advancement in wireless sensor networks (WSN) researchers in social network analysis (SNA) now have access to larger and more complex datasets that describe human interactions in the physical space. Studies in WSN thrive on accuracy and robustness whereas SNA operates on a higher level of data abstraction. Graph mining is a bridge between these two fields. This paper investigates two approaches to graph mining and compares their efficiency and appropriateness as the input systems for a social interaction analysis process.