Weaponized Tweets
Abstract
Beginning in Summer 2018, the Institute for Defense Analyses (IDA) devoted limited internal research and development funds to exploring foreign malign influence operations in social media. To date, we have focused on the Twitter activity of the Russian-backed Internet Research Agency (IRA) in the run-up to the 2016 American Presidential election. We used a data-driven, quantitative approach with human-in-the-loop, employing unsupervised Machine Learning techniques (Latent Dirichlet Allocation) and other data analytic methods. We focused on the evolution of the adversary activity, tracking how topic patterns in that activity changed over time from 2012 through 2018. Our results uncovered multiple distinct tactical phases of the IRA attack. The IRA first began targeting Americans in late 2014 to early 2015. Their English tweet topics grew more specific over time, more varied, more negative, and more polarizing. Their final tweet topic pattern first emerged around late 2015, one full year before the 2016 election.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jun 01, 2021
- Accession Number
- AD1210670
Entities
People
- Emily M. Parrish
- Jenny R. Holzer
- Shelley M. Cazares
Organizations
- Institute for Defense Analyses