Topic Modeling with Sentiment Evaluation for Analysis of Opinion Polarization

• In Event: 404 - Section on Social Psychology Paper Session. Computational Social Science and Studying Social Behavior

Abstract

In this paper we present a new text analysis technique, topic modeling with sentiment evaluation (TMSE), that can compare the degree of polarization of topics across multiple large text samples. We demonstrate TMSE by analyzing reactions to the Trayvon Martin controversy in spring 2012 by commenters on two partisan news websites. Based on studies of news media as an "outrage industry" and of political pundit inaccuracy, we predict that high-profile commentators (in this case Geraldo Rivera) will be more polarizing than other news personalities and topics. Results of the TMSE analysis support this prediction.

Authors

• Gabe Ignatow, University of North Texas

• Nicholas Evangelopoulos, University of North Texas

• Konstantinos Zougris, University of North Texas