Structural Micro Forces in Flickr Social Network

Abstract

Previous studies on network structure of large online social networks focused almost exclusively on exploring the global network descriptive statistics. Few of the studies that have researched how these networks evolve from the micro forces at the local level have fallen short of modeling reciprocity and different ways triangle closure. By focusing on the denser areas of the Flickr network (user groups) and with the help of recently extended biased net modeling framework, we specified and fitted models and estimated parameters for all possible purely structural dyadic and triadic network effects: reciprocity, transitivity, structural similarity, closure and cyclicality. Our results showed that the reciprocity is by far the most strongest force acting in the network, followed by transitivity, closure and structural similarity. Cyclicality has been, surprisingly, proven not to exist at all. Furthermore, we have found that the size of the groups corresponds negatively with the magnitude of each of the micro forces.

Keywords: Flickr, online networks, network structure, online social networks