With “fake news” being such a hot topic, ORAU and Penn State University are trying to answer this research question. The goal is to develop and validate a survey for measuring people’s susceptibility to fake news and refine a machine-learning tool designed to identify fake news.
Funded by the ORAU-Directed Research and Development program, this study is underway. Taking the supervised machine-learning approach, the team will collect a set of known fake news items that have been verified by crowdsourcing, trace how such misinformation was propagated on Twitter or Facebook and identify the users who have shared known misinformation at least once. These identified users form a group susceptible to misinformation, known in this study as positive labels. Identifiable user information will all be stricken from records once gathered, so there is no identifiable information associated with this end project.
In a parallel effort, the team will also identify a group of users who are not susceptible to misinformation. These people will be known as the negative labels—each received a piece of misinformation but has not shared it or has refuted it.
From these two sets of positive- and negative-labeled training data, the team will investigate how to build the best mathematical model to identify the gullibility of a given social media user using various statistical features. What is learned from this study, in combination with the other research ORAU is working on, can help ORAU employees and our customers better identify and guard against being susceptible to misinformation.
So, do you think you are gullible to fake news? Stay tuned for the results of this research to be announced in 2019.