Empowering Inclusive E-Deliberation through Stance Trees and Dialectic Trees

Abstract

In this paper, we show how inclusive e-deliberation can be enhanced through the use of artificial intelligence (AI). We focus on constructing "stance trees," which are hierarchical structures that organize polarized opinions by topic. Additionally, we introduce "dialectic trees," which go a step further by mapping arguments and counterarguments on specific issues. The proposed methodology integrates semantic information retrieval, topic modeling, stance prediction, and argument synthesis using generative AI, specifically large language models, to facilitate government-citizen interaction and public deliberation. By empowering citizens—including non-experts and minority groups—to contribute to decision-making processes, this research aims to foster more resilient social systems. The paper outlines current progress toward this goal.