The Problem
The United States has experienced increasing political polarization in recent years. Sentiment—both positive and negative—plays a vital role in shaping social dynamics by influencing how individuals and groups are perceived and treated. Hate speech, in particular, both drives and reflects societal change. When directed at groups, it is especially dangerous: if successfully “sold” to the public, all members of targeted groups become collectively blamed and subject to potential retribution, including violence.
The Model
We are introducing the Sentiment Analysis Model to systematically detect and interpret the intensity, direction, and group-based targets of emotionally challenged language in public discourse. By capturing both negative and positive sentiment, this model aims to provide early warnings of potential conflict, inform policy interventions, and support peacebuilding efforts through a deeper understanding of how language can both polarize and depolarize charged spaces. Read more →
The Method
This initiative uses automated and human coding to identify positive and negative speech by prominent people in media and politics against groups. It uses a Python-based tool to search Tweets and identifies potential units of analysis when positive or negative words and groups (from two dictionaries) are in close proximity.
Our Results
Results are presented via interactive data tables and charts to allow for easy segmentation and analysis.
