In this work, we propose a classification method for group vs. individual accounts on Twitter, based solely on communication network characteristics. While such a language-agnostic, network-based approach has been used in the past, this paper motivates the task from firmly established theories of human interactional constraints from cognitive science to sociology. Time, cognitive, and social role constraints limit the extent to which individuals can maintain social ties. These constraints are expressed in observable network metrics at the node (i.e. account) level which we identify and exploit for inferring group accounts.
This work was published in Lecture Notes in Computer Science and won the Best Paper Award at the International Conference on Social Computing, Behavioral-Cultural Modeling, and Prediction, Washington D.C.