Friend or Foe: Twitter Users under Magnification

In today's digital ecosystem, where people share vital information on a daily basis, it is imperative to identify security loopholes and vulnerability in social networks. Through identity resolution and disambiguation, information leakage and identity hijacking by malicious parties, can be reduced significantly. In this paper, we develop a simple model for successfully classifying Twitter users as suspicious and non-suspicious in their user activity. Our aim is to be able to find a concrete set of users that encompasses most users who misrepresent their identity on Twitter. Using user and tweet meta-data, we devised a mathematical model to tag users as listener, talker, hub, seed and absorber, and further conjugate the values generated by these equations, to identify suspicious and non-suspicious users in our 49,991-user dataset. This model of classification can be extended for integration with rigorous security mechanisms to identify true malicious users on Twitter and other online social platforms.