Detecting Covert Networks in Multilingual Groups: Evidence within a Virtual World

Abstract

This paper introduces an approach for the examination and organization of unstructured text to identify relationships between networks of individuals. This approach uses discourse analysis to identify information providers and recipients and determines the structure of covert organizations irrespective of the language that facilitate conversations between members. Then, this method applies social network analytics to determine the arrangement of a covert organization without any a priori knowledge of the network structure. This approach is tested and validated using communication data collected in a virtual world setting. Our analysis indicates that the proposed framework successfully detected the covert structure of three information networks, and their cliques, within an online gaming community during a simulation of a large-scale event.