Research on Cross-Platform Detection to Counter Malign Influence

Abstract

The most complicated malign influence campaigns, which are also among the greatest long-term national security threats, are those that move across multiple social platforms. The social media companies themselves have little capability in the area of cross-platform detection and analysis, which is one of the main reasons they partner with Graphika for their own defensive efforts. Each company has exceptional data and intelligence on what is happening on its own platform, but very few state actor campaigns are limited to a single platform where they can be detected and removed (deplatformed) entirely. In fact, as nation states have evolved their online operations, cross-platform networks have become part of the standard tradecraft. Sophisticated campaigns such as those emanating from China, Russia, and Iran typically have nexus on many platforms (see, for example, Graphikas Secondary Infektion investigation). As part of our current threat intelligence solutions, Graphika maps threat actor networks on individual platforms, and then our analysts and investigators manually trace connections as the campaigns move from one platform to another, eventually rolling them up for removal. Currently no technology exists to approach this problem in an automated way at scale.To combat this threat effectively overthe long term, research is needed to determine the feasibility of developing cross-platform capable network mappings and AI-enableddetection algorithms that can analyze volumes of data too large for any team of humans to reasonably process. Initial research in deep-learning graph models and network/language fingerprinting models shows the promise of this approach for generating very large network mappings of heterogeneous data sources. Upon success, a long-term capability could then be developed to provide high-quality, automated indicators and warnings (I&W) on sophisticated and malign campaigns at early stages before they can have significant impact.

Document Details

Document Type
DoD Grant Award
Publication Date
Jan 06, 2021
Source ID
N000142112106

Entities

People

  • Vladimir Barash

Organizations

  • Office of Naval Research
  • United States Navy

Tags

Fields of Study

  • Computer science

Readers

  • Agent-Based Social Robotics and Mobile-Assisted Learning in Virtual Environments.
  • Strategic Security Studies
  • Systems Analysis and Design

Technology Areas

  • AI & ML
  • AI & ML - Neural Networks