MArgE: Meshing Argumentative Evidence from Multiple Large Language Models for Justifiable Claim Verification

Ming Pok Ng, Junqi Jiang, Gabriel Freedman, Antonio Rago, Francesca Toni · 2025

Read MArgE: Meshing Argumentative Evidence from Multiple Large Language Models for Justifiable Claim Verification by Ming Pok Ng, Junqi Jiang, Gabriel Freedman, Antonio Rago, Francesca Toni (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2508-02584.txt.

Description

Ming Pok Ng and four co-authors present a 2025 arXiv paper on checking claims with several language models that must show their reasoning. Pooling models can curb single-model errors, but free-form debates leave no auditable trail. Their system, MArgE, forces each model to lay its evidence out as a tree of arguments, built with tools from formal argumentation theory.

The trees can be inspected step by step, so the final verdict on a claim carries a faithful justification. In tests the system beat single models, including a small commercial one, earlier argument-based models, and unstructured debate setups. The gains held across models of modest size. The authors pitch structured argument as the missing discipline in multi-model systems: not just more voices, but voices that must answer to a shared logic.

Catalog record

CREATORMing Pok Ng, Junqi Jiang, Gabriel Freedman, Antonio Rago, Francesca Toni
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE50 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2508-02584

Browse the full library · About the Tower of Babel