- This event has passed.
Alexandre Bovet (UNamur – UCL)
March 1, 2018 @ 13:00 - 14:00
Title: Opinion dynamics and fake versus traditional news influence in Twitter
Abstract: We investigate the opinion of Twitter users during the 2016 US elections using a large scale dataset of more than 170 millions tweets. We develop a method to infer the opinion of Twitter users regarding the candidates by using a combination of natural language processing of the tweet contents, machine learning classification and analysis of the hashtags co-occurrence network. We study the temporal social networks formed by the interactions among millions of Twitter users and infer the support of each user to the presidential candidates. The resulting Twitter opinion trend follows the New York Times National Polling Average, which represents an aggregate of hundreds of independent traditional polls, with remarkable accuracy.
Going beyond the daily opinion analysis and analyzing the level of activity, the repartition of the supporters between the strongly connected giant component and the rest of the network and the daily fluctuations in the number of users reveal a clear dichotomy between the behavior of supporters of each candidate. Although Clinton supporters are the majority in Twitter, Trump supporters are generally more active and more constant in their support, while Clinton supporters are less active and show their support only occasionally.
To understand the role of information diffusion on Twitter opinion dynamics, we consider tweets containing URLs directing to news outlet websites. In particular, we compare websites known to diffuse fake news compared to traditional, fact-based, news outlets. We find that 29% of the tweets linking to news outlets points to websites containing fake or extremely biased news. Analyzing the information diffusion networks, we find that user diffusing fake news form more connected and less heterogeneous networks than users in traditional news diffusion networks. While influencers of traditional news outlets are journalists and public figures with verified Twitter accounts, most influencers of fake news and extremely biased websites are unknown users or users with deleted Twitter accounts. Finally, an analysis of the activity dynamics of influencers reveals that influencers of tradition news are driving the most part of Twitter while fake news influencers are, in fact, mostly driven by the activity of Trump supporters.