Coordinated Behavior at Scale: A Minimalist Approach to Detecting Information Operations

Description

This work presents a computationally lightweight and language-agnostic methodology for detecting coordinated influence campaigns at scale on social media using only minimal, widely available metadata (timestamps, user IDs, hashtags, mentions, and interaction references). By focusing on behavioral synchronization within short temporal windows rather than platform-specific parameters, the proposed pipeline constructs Highly Coordinated Networks (HCNs) and filters accidental co-occurrences using an FSA-V backbone pruning technique, a temporal null model, and temporal persistence checks. Evaluated across 16 benchmark information-operation datasets encompassing 15 campaigns, 11 countries, and 6 languages, the approach demonstrates strong generalization, consistently achieving F1 scores above 0.7 and identifying over 80% of target posts while executing in minutes. The method also extracts key campaign-specific actions to enable downstream narrative analysis and agentic AI integration.

Period	17 Dec 2025
Event title	11th Belgian Network Research Meeting
Event type	Conference
Location	Ghent , BelgiumShow on map
Degree of Recognition	National