Investigating the Use of Bots to Spread Fake News in Social Media

Fake news is not new. However, the phenomenal growth of online social media coupled with ease in publishing unverified content and click based advertisement revenue, have increased the use of fake news to drive discussions. Though often considered innocuous, it can have high social cost. For example, Allcott and Gentzkow [1] in their study on 2016 presidential election, find around 38 million shares of fake news on Facebook, with 30 million pro-Trump and 7.6 million pro-Clinton shares. Parkinson [7] reported that the inactivity of social-media companies in removing fake-news might have contributed to the results of the 2016 US presidential election. Either for spreading ideologies or for making money, the goal of fake news creators is to spread their message rapidly, so the likely use of bots (or human assisted bots) in the process is hard to dismiss. However, most existing research on fake-news have considered the origin [8], the motivations [1] and the impact of fake news [7], but not the use of bots. In our investigation, we have found social bots [6] that are actively being used on Twitter to fool the content promotion algorithms and spread specific agenda. In this research, we use network analysis to understand how social botnets are used for the faster dissemination of fake news, and try to find the motivations behind such sharing by jointly analyzing the bots behavior and the change of network structure over time. Besides, we also investigate the syntactic characteristic of social-media posts related to fake-news on Twitter, including the use of hashtags, user-mentions and sentiment that make fake news more appealing to particular groups.