Monitoring Online Discourse for Election Motivated Violence

• In Event: Media and Mobilization to Extreme Violence

Abstract

Increasingly, access to news is shifting from traditional sources (e.g. print and television) to online media platforms that included elements of social engagement. For news sites like New York Post, Washington Times, Common Dreams and many others, the combination of articles with their corresponding user comments provides a unique insight into the broader discourse regarding highly politicized topics, some of which have translated into hate, threats, and the actualization of physical violence. Prior and ongoing examples include the targeting of the LGBTQI+ and Jewish communities, as well as immigrants along the US-Mexico boarder. With a focus on election-related violence, we have developed a real-time monitoring system that tracks online discourse (both articles and user comments) to better characterize, understand, and predict the risk of target violence towards individuals, groups, and institutions. Applying natural language processing techniques on thousands of articles over a 6 month period we demonstrate that online discourse can be matched to known cases of election-related violence towards government officials.

Authors

• Adan Vela, University of Central Florida

• Steven Windisch, University of Arkansas