Predicting Disease Outbreaks Using Social Media: Finding Trustworthy Users

The use of Internet data sources, in particular social media, for biosurveillance has gained attention and credibility in recent years. Finding related and reliable posts on social media is key to performing successful biosurveillance utilizing social media data. While researchers have implemented various approaches to filter and rank social media posts, the fact that these posts are inherently related by the credibility of the poster (i.e., social media user) remains overlooked. We propose six trust filters to filter and rank trustworthy social media users, as opposed to concentrating on isolated posts. We present a novel biosurveillance application that gathers social media data related to a bio-event, processes the data to find the most trustworthy users and hence their trustworthy posts, and feeds these posts to other biosurveillance applications, including our own. We further present preliminary experiments to evaluate the effectiveness of the proposed filters and discuss future improvements. Our work paves the way for collecting more reliable social media data to improve biosurveillance applications.