Online Program

Return to main conference page
Thursday, May 30
Data Science Techologies
Practice and Applications
Data Science Applications E-Posters, I
Thu, May 30, 3:00 PM - 4:00 PM
Grand Ballroom Foyer

Analyzing Influence of Social Media Through Twitter (306368)


William Boettcher, North Carolina State University 
*Dhrubajyoti Ghosh, North Carolina State University 
Rob Johnston, North Carolina State University 
Michele Kolb, North Carolina State University 
Soumendra Lahiri, North Carolina State University 
James Robertson, North Carolina State University 

Keywords: Twitter, Influence, Social Media, Influencers

The influence of social media on political campaigns has started gaining significant attention since the 2016 US Presidential Elections. There are a plethora of works devoted to retrospectively identifying influence efforts by malign state actors that were deployed through common social media platforms—Facebook, Twitter, Instagram, etc. Less attention has been devoted to the underlying general mechanisms for generating influence on these platforms and the degree to which this influence translates into electoral outcomes. This work focuses on these lesser explored aspects of social media influence, in a case study of the recent Irish Abortion Referendum using a dataset made available by Twitter. A time series analysis of dynamic networks was conducted to analyze the mechanisms associated with persuasion propagating from core "Influencers" and the degree to which "Influence" varied across messages advocated. We also demonstrate a relatively rigorous forecasting model of the referendum, which produced fairly efficient predictions, duly noting that other electoral campaigns may not produce such clear parallels between the realm of Twitter exchange and the real world.