From Passive Response to Proactive Correction: Enhancing LLM Robustness Against Input Fact Perturbations

2026-08-26

AI Summary

本文针对输入事实扰动导致大型语言模型产生错误回答的问题，提出DEDUCE框架，通过检测、策略制定和纠正三个阶段增强模型的鲁棒性和准确性。

Abstract

Large language models (LLMs) frequently produce confident yet factually incorrect responses when user inputs contain misleading premises, a phenomenon we attribute to fact perturbations in the input. Existing approaches to hallucination mitigation typically assume reliable user inputs, overlooking how such factual errors can actively mislead model reasoning. To address this vulnerability, we propose DEDUCE, a three-stage framework that transforms LLMs from passive responders into proactive error correctors. DEDUCE operates in three stages: (1) detect errors through fine-grained fact extraction and verification; (2) devise correction strategies via multi perspective deliberation; and (3) correct misconceptions while delivering reliable answers. We also present MisFactQA, a dataset containing factual errors of varying degrees, and propose new metrics for evaluating model robustness. Experiments on TruthfulQA, FalseQA, and our MisFactQA benchmark demonstrate that DEDUCE significantly improves both accuracy and error correction capability. Consistent gains across Qwen, LLaMA, and Gemma families confirm its effectiveness and scalability.

Problem

• large language models
• fact perturbations
• hallucination mitigation
• model reasoning

Innovation

• DEDUCE
• fact extraction and verification
• multi perspective deliberation
• MisFactQA