A statistical approach for social network change detection: an ERGM based framework
S. Golshid Sharifnia, Abbas Saghaei
Abstract
Social networks have an important role in today’s lifestyle and detecting any changes in their structure could be vital for social systems. Change detection in a social process could be complicated because of stochastic and complex manners of its component’s which are humans. In this article, a novel approach is proposed which model and analyze social networks’ structure. For this purpose, the proposed method combines the nodal attributes with structural tendencies of ERGMs to find the best fitting model which can properly define humans’ characteristics in the observed network. Then, to detect any changes in the proposed model Hotelling T2 and MEWMA control charts are employed. Experimental simulation study and data analysis demonstrated the efficiency of the proposed technique to detect changes and its sensitivity in finding anomalies.