Graduate Seminar: Understanding Social Media Credibility: Challenges for Social Computing

Event Date: 
Thu, 2017-02-16 09:30 - 10:45

Location: McBryde 655

Speaker: Tanushree Mitra, Georgia Tech

Abstract:
Today, social media provide the means by which billions of people produce, consume and distribute information. This is often empowering, but can also be disruptive. What if someone says something false? In today’s talk, I will focus on this line of inquiry.

Modern online social networks are neutral towards the credibility of information. Simply put, they transmit both credible and less credible information. In this talk, I will focus on addressing this specific challenge — the challenge of assessing the credibility of social media information in the absence of traditional gatekeepers. First, I will present the development of the first large-scale, systematic social media credibility corpus, called CREDBANK – a corpus comprising 66M tweets nested in 1,377 real-world events. Second, by discussing the analysis of CREDBANK, I will show that temporal and linguistic regularities can differentiate credible and non-credible information. Finally, I will conclude with a preview of a new line of work addressing additional challenges that arise due to lack of information gatekeeping in social computing systems.