How “VKontakte” fake accounts influence the social network of users

Abstract

In this paper, the problem of fake accounts in online social networks is addressed through the lens of resulting misstatements of the structure of network interactions between users. The study of a network as a social space becomes difficult because of additional noise created by fakes. The aim of the present paper is to assess the impact of fake accounts on the characteristics of local friendship networks between users of VKontakte website in Izhevsk (Russia). The authors highlight key characteristics recognizing a fake account and present experience of the design of classifier (based on random forest algorithm) to determine the authenticity of an account. Comparison of the VKontakte network topology before and after removing the fake accounts from it shows what specific network metrics are affected by the presence of fake profiles. It was found that as the fakes are being excluded the less integrated members lose contact with most part of the network while the number of its components increases. “Fakes” serve as strong link concentrators distributed throughout the network and these fakes overestimate observed levels of assortativity and transitivity.