Home / Projects / RADII: Reticular Analysis of Discourse for Influence Indicators

RADII: Reticular Analysis of Discourse for Influence Indicators

The project seeks to develop novel AI and NLP techniques for building a highly automated, user-driven toolkit that identifies Influence Indicators (IIs) and enables intelligence analysts to map the influence terrain online and discover dynamic shifts in it. FIU, in collaboration with Smart Information Flow Technologies (SIFT), the Institute for Human and Machine Cognition (IHMC), Rensselaer Polytechnic Institute (RPI), and Stony Brook University (SBU) will develop RADII to continually orchestrate a set of innovative Influence Indicator Detectors (IIDs) for detecting linguistic behaviors associated with persuasion attempts in social media. In accord with analyst needs, direction and intuitions, RADII will automatically organize low-level data into actionable information to detect and understand influence campaigns. RADII will process mass social media data rapidly and insightfully for IIs that occur at high-, mid- and low-levels and across sources, and then estimates its confidence in context by propagating local values from each detector. This flexibility greatly benefits U.S. operations in the information environment as well as corporate security efforts, while keeping the analyst in charge of, but not overwhelmed by, analysis efforts. RADII is part of the DARPA INCAS program.

Dates Active: June 1, 2018 — May 31, 2023


Defense Advanced Research Projects Agency (DARPA; funder)
Smart Information Flow Technologies, Inc. (SIFT; collaborator)
Institute for Human and Machine Cognition (IHMC; collaborator)
Rensselaer Polytechnic Institute (RPI; collaborator)
Stony Brook University (SBU; collaborator)