How Gullible Are You? Predicting Susceptibility to Fake News

Shen, Tracy Jia; Cowell, Robert; Gupta, Aditi; Le, Thai; Yadav, Amulya; Lee, Dongwon

In this research, we hypothesize that some social users are more gullible to fake news than others, and accordingly investigate on the susceptibility of users to fake news--i.e., how to identify susceptible users, what are their characteristics, and if one can build a prediction model.Building on the crowdsourced annotations of 5 types of susceptible users in Twitter, we found out that: (1) susceptible users are correlated with a combination of user, network, and content features; (2) one can build a reasonably accurate prediction model with 0.82 in AUC-ROC for the multinomial classification task; and (3) there exists a correlation between the dominant susceptibility level of center nodes and that of the entire network.