IOHunter: Graph Foundation Model to Uncover Online Information Operations

  • 2025-03-03 15:32:17
  • Marco Minici, Luca Luceri, Francesco Fabbri, Emilio Ferrara
  • 0

Abstract

Social media platforms have become vital spaces for public discourse, servingas modern agor\`as where a wide range of voices influence societal narratives.However, their open nature also makes them vulnerable to exploitation bymalicious actors, including state-sponsored entities, who can conductinformation operations (IOs) to manipulate public opinion. The spread ofmisinformation, false news, and misleading claims threatens democraticprocesses and societal cohesion, making it crucial to develop methods for thetimely detection of inauthentic activity to protect the integrity of onlinediscourse. In this work, we introduce a methodology designed to identify usersorchestrating information operations, a.k.a. IO drivers, across variousinfluence campaigns. Our framework, named IOHunter, leverages the combinedstrengths of Language Models and Graph Neural Networks to improvegeneralization in supervised, scarcely-supervised, and cross-IO contexts. Ourapproach achieves state-of-the-art performance across multiple sets of IOsoriginating from six countries, significantly surpassing existing approaches.This research marks a step toward developing Graph Foundation Modelsspecifically tailored for the task of IO detection on social media platforms.

 

Quick Read (beta)

loading the full paper ...