Head of AI Technology, AllSides
Computer Scientist building the bridge between human discourse and machine reasoning. Research leverages Natural Language Processing to deconstruct argumentation patterns within societal dialogues, creating the foundation for more robust, explainable, and aligned AI systems. This work is critical for developing autonomous agents that can reason soundly about complex, human-centric problems and contribute to a healthier public sphere.
Computational tools for analyzing political discourse to detect logical fallacies, reduce echo chamber effects, and bridge representation gaps in democratic processes. Framework enables systematic comparison of public opinion with legislative debates at unprecedented scale.
Bias detection systems deployed in production to create more aligned AI that understands human reasoning patterns and provides transparent explanations. Argumentative bias detection methods enable humans to challenge AI conclusions effectively.
Discourse analysis tools to identify manipulation patterns in information supply chains and guide counter-disinformation strategies during critical events. Research published guides systematic understanding of what is being argued, by whom, and with what evidence.
Introduces DALiSM, a data-driven argument-centric framework to analyze discourse dynamics in diverse and multi-party spaces at scale, from formal legislative proceedings to online discussion forums. Transforms how we analyze deliberative dynamics across formal legislative proceedings and online forums.
A comprehensive framework for understanding arguments and debate in online forums, featuring methods for detecting argument topics, describing argument structure, and quantifying content diversity through clustering algorithms. Cited by 6 works.
A comprehensive framework that detects the existence, topic, and stance of arguments across contexts, leveraging fine-tuning and prompt-engineering of Large Language Models. Achieves 80-86% accuracy in identifying argumentative structures. Cited by 4 works.
Investigates over 130,000 Telegram messages and 750 news articles to assess Russian information supply chains, revealing neutral language patterns and information laundering networks in war reporting.
A systematic framework to understand the intertwined change of sentiment and stance due to real-world events in online discussions, analyzing 7.5 million posts across 4chan, Reddit, and Parler during the 2020 US Election. Cited by 8 works.
Production-ready platforms used by researchers at 5+ universities
Comprehensive argument mining framework that detects existence, topic, and stance of arguments across contexts. Leverages fine-tuning and prompt-engineering of Large Language Models for analyzing discourse in various text formats.
Web-based platform for analyzing and visualizing discourse dynamics. Processes text data from Congressional hearings, Reddit threads, and other sources with CSV upload and API retrieval capabilities.