Adaptive Collaboration of Arena-Based Argumentative LLMs for Explainable and Contestable Legal Reasoning

2026-02-21

AI Summary

This work addresses the lack of verifiable and contestable structured explanations in current large language models for legal reasoning, which hinders auditability and user intervention. The authors propose a neuro-symbolic hybrid framework that, for the first time, integrates adaptive multi-agent collaboration with a formal argumentation system. Built upon an arena-based Quantitative Bipolar Argumentation Framework (A-QBAF), the approach dynamically generates, adjudicates, and refines legal arguments. It introduces mechanisms for conflict resolution, uncertainty-aware escalation, and a human-in-the-loop contestable workflow, thereby enabling transparent and intervenable reasoning. Evaluated on the LegalBench benchmark, the model—implemented within the Gemini-2.5-Flash architecture—significantly outperforms strong baselines, achieving both high predictive accuracy and structured interpretability.

Technology Category

• Knowledge Representation and Reasoning: Argumentation
• Multiagent Systems: Agreement, Argumentation & Negotiation
• Natural Language Processing: (Large) Language Models

Application Category

• Search and Retrieval-Augmented AI: Large language models for search
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Social Networks and Social Media: Generative AI / large language models and their impact on social systems

Abstract

Legal reasoning requires not only high accuracy but also the ability to justify decisions through verifiable and contestable arguments. However, existing Large Language Model (LLM) approaches, such as Chain-of-Thought (CoT) and Retrieval-Augmented Generation (RAG), often produce unstructured explanations that lack a formal mechanism for verification or user intervention. To address this limitation, we propose Adaptive Collaboration of Argumentative LLMs (ACAL), a neuro-symbolic framework that integrates adaptive multi-agent collaboration with an Arena-based Quantitative Bipolar Argumentation Framework (A-QBAF). ACAL dynamically deploys expert agent teams to construct arguments, employs a clash resolution mechanism to adjudicate conflicting claims, and utilizes uncertainty-aware escalation for borderline cases. Crucially, our framework supports a Human-in-the-Loop (HITL) contestability workflow, enabling users to directly audit and modify the underlying reasoning graph to influence the final judgment. Empirical evaluations on the LegalBench benchmark demonstrate that ACAL outperforms strong baselines across Gemini-2.5-Flash-Lite and Gemini-2.5-Flash architectures, effectively balancing efficient predictive performance with structured transparency and contestability. Our implementation is available at: https://github.com/loc110504/ACAL.

Problem

• legal reasoning
• explainability
• contestability
• argumentation
• Large Language Models

Innovation

• neuro-symbolic framework
• argumentative multi-agent collaboration
• quantitative bipolar argumentation
• human-in-the-loop contestability
• adaptive legal reasoning