This class will bring together readings and discussions of political theory and empirical studies of public opinion with hands-on explorations of natural language processing tools and methods — from classic methods to modern deep neural modeling — for measuring public opinion and systematically understanding other aspects of public thought. The class will aim to understand what public opinion is, how it has been measured and conceptualized in the past, to imagining and attempting to implement how a broad range of public thought and experience can be analyzed and measured in the future, given the affordances of contemporary information processing technology and abundance of data. Our goal is to design and advance new strategies for measuring public opinion, paying particular attention to the need to understand perspectives as well as gauge preferences.
This class will be offered jointly at University of Wisconsin-Madison and MIT with students from both universities in a single class. We expect an interdisciplinary mix of students, creating an opportunity for group projects that span social science, AI/machine learning, journalism and design.