MAD-Fact: A Multi-Agent Debate Framework for Long-Form Factuality Evaluation in LLMs

2025/10/27 by Ning, Yucheng, Lin, Xixun, Fang, Fang +1
#Artificial Intelligence (cs.AI) #Computation and Language (cs.CL) #FOS: Computer and information sciences

paper · doi:10.48550/arxiv.2510.22967

Abstract

The widespread adoption of Large Language Models (LLMs) raises critical concerns about the factual accuracy of their outputs, especially in high-risk domains such as biomedicine, law, and education. Existing evaluation methods for short texts often fail on long-form content due to complex reasoning chains, intertwined perspectives, and cumulative information. To address this, we propose a systematic approach integrating large-scale long-form datasets, multi-agent verification mechanisms, and weighted evaluation metrics. We construct LongHalluQA, a Chinese long-form factuality dataset; and develop MAD-Fact, a debate-based multi-agent verification system. We introduce a fact importance hierarchy to capture the varying significance of claims in long-form texts. Experiments on two benchmarks show that larger LLMs generally maintain higher factual consistency, while domestic models excel on Chinese content. Our work provides a structured framework for evaluating and enhancing factual reliability in long-form LLM outputs, guiding their safe deployment in sensitive domains.

Citations

• LLM-based Agents Suffer from Hallucinations: A Survey of Taxonomy, Methods, and Directions
• M-MAD: Multidimensional Multi-Agent Debate for Advanced Machine Translation Evaluation
• JudgeBlender: Ensembling Judgments for Automatic Relevance Assessment
• LLMs for Time Series: an Application for Single Stocks and Statistical Arbitrage
• Truth or Mirage? Towards End-to-End Factuality Evaluation with LLM-Oasis
• From Generation to Judgment: Opportunities and Challenges of LLM-as-a-judge
• Chinese SimpleQA: A Chinese Factuality Evaluation for Large Language Models
• FactBench: A Dynamic Benchmark for In-the-Wild Language Model Factuality Evaluation
• FIRE: Fact-checking with Iterative Retrieval and Verification
• Self-Pluralising Culture Alignment for Large Language Models
• Enhancing Agricultural Machinery Management through Advanced LLM Integration
• On Large Language Models in National Security Applications
• VERISCORE: Evaluating the factuality of verifiable claims in long-form text generation
• Towards Detecting LLMs Hallucination via Markov Chain-based Multi-agent Debate Framework
• Hallucination of Multimodal Large Language Models: A Survey
• RAGAR, Your Falsehood Radar: RAG-Augmented Reasoning for Political Fact-Checking using Multimodal Large Language Models
• Long-form factuality in large language models
• Improving Retrieval for RAG based Question Answering Models on Financial Documents
• FinTral: A Family of GPT-4 Level Multimodal Financial Large Language Models
• LLM Multi-Agent Systems: Challenges and Open Problems
• Large Language Model based Multi-Agents: A Survey of Progress and Challenges
• Large Language Models for Generative Information Extraction: A Survey
• Advancing Transformer Architecture in Long-Context Large Language Models: A Comprehensive Survey
• Factcheck-Bench: Fine-Grained Evaluation Benchmark for Automatic Fact-checkers
• A Survey on Hallucination in Large Language Models: Principles, Taxonomy, Challenges, and Open Questions
• Exploiting User Comments for Early Detection of Fake News Prior to Users' Commenting
• FreshLLMs: Refreshing Large Language Models with Search Engine Augmentation
• Evaluating Hallucinations in Chinese Large Language Models
• FELM: Benchmarking Factuality Evaluation of Large Language Models
• Towards LLM-based Fact Verification on News Claims with a Hierarchical Step-by-Step Prompting Method
• The Rise and Potential of Large Language Model Based Agents: A Survey
• A Survey of Hallucination in Large Foundation Models
• Siren's Song in the AI Ocean: A Survey on Hallucination in Large Language Models
• 🧜Siren’s Song in the AI Ocean: A Survey on Hallucination in Large Language Models
• ChatEval: Towards Better LLM-based Evaluators through Multi-Agent Debate
• Med-HALT: Medical Domain Hallucination Test for Large Language Models
• FacTool: Factuality Detection in Generative AI -- A Tool Augmented Framework for Multi-Task and Multi-Domain Scenarios
• Chatlaw: A Multi-Agent Collaborative Legal Assistant with Knowledge Graph Enhanced Mixture-of-Experts Large Language Model
• Self-Checker: Plug-and-Play Modules for Fact-Checking with Large Language Models
• Enabling Large Language Models to Generate Text with Citations
• Improving Factuality and Reasoning in Language Models through Multiagent Debate
• FActScore: Fine-grained Atomic Evaluation of Factual Precision in Long Form Text Generation
• HaluEval: A Large-Scale Hallucination Evaluation Benchmark for Large Language Models
• A Survey of Large Language Models
• SelfCheckGPT: Zero-Resource Black-Box Hallucination Detection for Generative Large Language Models
• GPT-4 Technical Report
• LLaMA: Open and Efficient Foundation Language Models
• When Not to Trust Language Models: Investigating Effectiveness of Parametric and Non-Parametric Memories
• RARR: Researching and Revising What Language Models Say, Using Language Models
• Factuality Enhanced Language Models for Open-Ended Text Generation
• OPT: Open Pre-trained Transformer Language Models
• BNAI, NO-TOKEN, and MIND-UNITY: Pillars of a Systemic Revolution in Artificial Intelligence
• TruthfulQA: Measuring How Models Mimic Human Falsehoods
• A Survey on Automated Fact-Checking
• Language Models are Few-Shot Learners
• A Survey on LLM-based Multi-Agent System: Recent Advances and New Frontiers in Application
• DebUnc: Improving Large Language Model Agent Communication With Uncertainty Metrics

Related

• Can LLMs Produce Faithful Explanations For Fact-checking? Towards Faithful Explainable Fact-Checking via Multi-Agent Debate
• Hallucination to Truth: A Review of Fact-Checking and Factuality Evaluation in Large Language Models
• Long-form factuality in large language models
• FaStfact: Faster, Stronger Long-Form Factuality Evaluations in LLMs
• Are Large Language Models Reliable Judges? A Study on the Factuality Evaluation Capabilities of LLMs