Technical ReportCS 418 · UIC Fall 2025·2025

Coordinated Amplification and Misinformation Detection in Global YouTube Conflict Narratives

Archit RathodSrinath GaneshVishaal DayashankerHarsh ShelkeVignesh Pathak
abstract

YouTube serves as a major conduit for viral, multilingual political narratives, particularly during global conflicts. This project investigates coordinated amplification patterns and misinformation detection in YouTube content related to the Russia-Ukraine conflict. We analyzed approximately 5.9 million comments across 440,772 videos from 1,561 channels using a multi-method approach combining network science, anomaly detection, and natural language processing. Our findings validate three core hypotheses: (1) misinformation is amplified by highly interconnected channel and commenter clusters, (2) periods of intense real-world conflict correlate with statistically significant engagement anomalies, and (3) narratives evolve predictably over time in alignment with external war events. The project demonstrates the effectiveness of combining PageRank centrality analysis, Isolation Forest anomaly detection, and BERTopic modeling for detecting coordinated information campaigns at scale.

keywords
Misinformation DetectionYouTubeNetwork SciencePageRankBERTopicAnomaly DetectionNLPRussia-Ukraine Conflict