Modeling polarization in public opinion through LLM-synthesized arguments and stance trees

Existing e-deliberation and argument mining systems struggle to synthesize coherent arguments from noisy social media data and reveal polarization patterns in scalable, interpretable forms. To address these limitations, we present a methodology that leverages Large Language Models (LLMs) to model polarization in public opinion through structured stance trees , which organize collective opinions by topic and stance to support inclusive e-deliberation. LLMs play a central role by generating synthesized arguments that capture the reasoning underlying cohesive opinion clusters, helping reveal polarization patterns in informal online discourse and transforming it into interpretable argumentative forms. Unlike previous work in argument mining, which primarily focuses on identifying and classifying existing argumentative components such as claims and premises, our framework emphasizes argument synthesis as a generative process. We introduce a dataset that links clusters of related opinions with their corresponding LLM-synthesized arguments, annotated by human experts for coherence, relevance, and argumentative quality. The experimental study evaluates the quality of these LLM-synthesized arguments using both human experts and LLMs as judges, examining the degree of consensus between human and automated assessments. We compare three open-source LLMs using both evaluation approaches. This resource and methodology provide a foundation for advancing research in generative argumentation and for developing deliberative tools that help policymakers and citizens better understand public reasoning and contrasting viewpoints.