Abstract

In addition to the lively discussion by genuine social media users, there is often an effort by duplicitous users to sway the conversation in a particular direction. Traditionally, these efforts are carried out by bots and crowdturfers who are programmed or recruited in mass to spread posts about a product, candidate or cause. In this work, we propose to study a new class of duplicitous user,“shills.” Shills are professionals who carry out targeted replies to users with opposing views, often at the pleasure of a campaign. Shills are real users hired by an organization in order to promote an idea or agenda by targeted combating of any opposing opinions. We investigate whether we can effectively characterize shills and further verify if it is possible to automatically detect them. This work represents the first effort in defining social media shills, characterizing them, and suggesting a data mining framework to explore this new, challenging problem.

Date

2017

Authors

Fred Morstatter, Robert P Trevino, Huan Liu

Journal

SBP