Identifying Online Calls for Election Related Violence

• In Event: Media and Mobilization to Extreme Violence

Abstract

In the lead-up to and following the 2020 Presidential Election, concerns of electoral fraud spread throughout traditional and social media outlets. What followed were isolated efforts by individuals and targeted campaigns by groups that sought to gather information, advocate positions, and seek to discover “the truth” from lawmakers, government officials, and regular electoral staff. In many cases, the line between political action and election-motivated violent extremism dissolved, whereby officials reported receiving threats of violence via phone, mail, email, and social media platforms. In an effort to better understand the level of acceptance and preponderance of political and electoral violence within unmoderated online communities we examine discourse on the social media platform Telegram. Focusing on user comments to posts and utilizing natural language processing techniques we develop classifiers to identify malevolent comments that advocate or call for punishment, harm, or violence to lawmakers and government officials. Using the classifier we are able to characterize the prevalence and nature of calls for harm and violence overtime, across platforms, and online communities.

Authors

• Adan Vela, University of Central Florida

• Steven Windisch, University of Arkansas