Misinformation on social media poses profound societal challenges, yet meaningful research and response efforts remain constrained by a lack of reliable tion methods that empower researchers to identify and analyze deceptive content. This thesis presents a scalable, generalizable machine learning system for misinformation detection across diverse social media platforms and topic domains. To build an effective probabilistic model, we introduce novel metrics derived from compre hensive studies in claim detection, formal logical fallacy identification, and author reliability propagation. These innovations are supported by several original, human-annotated datasets collected from Twitter, Telegram, and Reddit, covering critical topics such as elections, public health, and armed conflict. By providing a flexible and robust analytical lens, this work advances large-scale social cybersecurity research and can be integrated with existing tools to enhance the study of information spaces. To demonstrate the practical utility of our system, we conclude with a multicase study analysis and identify key directions for future research to further improve misinformation detection and understanding

150 pages

Thesis Committee:
Kathleen M. Carley (Chair)
Hong Shen
Brandy Aven
Dave Beskow (United States Military Academy)

Nicolas Christin, Head, Software and Societal Systems Department
Martial Hebert, Dean, School of Computer Science

Return to: SCS Technical Report Collection
School of Computer Science

This page maintained by reports@cs.cmu.edu