Using Machine Learning to Infer Real-world Political Attitudes and Behaviors from Social Media Data

Abstract

Social media has become a primary means of communication and personal expression for many, and digital trace data from social media platforms can contain rich and extensive archives of individuals’ attitudes, beliefs, and actions. But even though these data are increasingly plentiful and available to social science researchers, the process of extracting meaningful measures of individual-level attributes from large collections of social media data is nontrivial. In this talk, Computational Social Scientist, Sam Bestvater, will draw from his research on political engagement in online spaces and its impacts on real-world behaviors to discuss how machine learning algorithms can be used to analyze large amounts of social media data and extract insights into political attitudes and activities. Along the way, the talk will introduce several recent innovations in natural language processing and computer vision, and will discuss some potential challenges and limitations of using these tools and data sources for political research, as well as ethical considerations that should be taken into account.