The study seeks to understand the dissemination of fake news among Twitter users during the two weeks leading up to the 2016 US Presidential elections. An in depth study was conducted by identifying tweets that contained known fake news domains. With this data the study explored dimensions such as temporal tweet activity, likelihood of account automation, tweet volume, temporal persistence of widely shared urls, and candidate affinity.
Highlights:
- The study found that fake news urls seemed to originate from fake news sites, the most popular of which are run from accounts which display bot-like behaviour.
- The most popular urls propagated in the two weeks building up to the 2016 US presidential elections were predominantly based on "Anti-Clinton" and "Pro-Trump" sentiment.
- According to the main author, the website 'BotorNot' should be used with caution when establishing conclusions about an account's automation. The results of 'BotorNot' among users in the study's corpus who appeared bot-like and real word-users showed similar overall scores.