Elite Network Events from Authoritarian News Sources

• In Event: Methodological Developments in Text-as-Data

Abstract

Elite relationships and network dynamics are core components of authoritarian regime stability. They can help explain regime outcomes such as coups and purges. Yet it can be difficult to detect how various authoritarian elites interact with one another due to the relatively closed information systems. In this project, we demonstrate how to track interactions between authoritarian elites using a newly compiled corpus of news articles from media and government sources. We scrape image and text data from online Russian news sources and then use natural language processing and computer vision tools to extract event data from those sources. We define elites as prominent public figures from a range of spheres, including the government, military, and religious organizations. Importantly, we extract both positive and negative interactions between elites and consider many different types of interactions (e.g. instances of one elite praising another as well as instances of one elite criticizing another). The extracted ties allow us to construct signed networks of elite interaction over time.

Authors

• Nora Webb Williams, University of Illinois at Urbana-Champaign

• Tianhong Yin, University of Illinois at Urbana-Champaign

• Emily Sallenback, Princeton University

• Michael Gabbay, University of Washington