---
title: "从被动应答到主动纠错：DEDUCE 框架增强大模型对输入事实扰动的鲁棒性"
source: "arXiv AI/CL/LG（论文）"
source_tier: "t2"
url: "https://arxiv.org/abs/2608.25894v1"
permalink: "https://feed.astralor.com/items/itm_7e154ed693460246899c1dcb"
published: "2026-08-26T15:13:50.000Z"
score: 71
channel: "ai"
tags: ["DEDUCE","MisFactQA","大语言模型","幻觉缓解","输入事实扰动"]
---

# 从被动应答到主动纠错：DEDUCE 框架增强大模型对输入事实扰动的鲁棒性

## AI 点评

> 这篇论文值得 AI 产品团队关注，因为用户输入中的事实性错误比想象中常见，而多数模型只会顺着错误信息作答。DEDUCE 把“纠错”作为显式环节，而不是事后兜底，这对客服、搜索、金融问答等场景尤其有价值。如果后续开源代码和模型，落地门槛会更低。

## 速览

大语言模型在用户输入包含误导性事实时，常会自信地给出错误回答。现有幻觉缓解方法大多假设用户输入可靠，忽视了这类事实扰动对推理的诱导。本文提出 DEDUCE 三阶段框架，通过细粒度事实提取与验证检测错误、多视角推演制定纠正策略，并最终纠正误解的同时给出可靠答案。他们还构建了 MisFactQA 数据集并提出新评测指标，在 TruthfulQA、FalseQA 及 MisFactQA 上验证了 DEDUCE 能显著提升准确率与纠错能力，且对 Qwen、LLaMA、Gemma 均有效。

## 关键点

- 识别并定义“输入事实扰动”问题，指出现有幻觉缓解方法的盲区。
- 提出 DEDUCE 三阶段框架：检测错误、多视角推演、纠正并给出可靠答案。
- 发布 MisFactQA 数据集及新的鲁棒性评估指标。
- 在 Qwen、LLaMA、Gemma 等模型上验证了显著且一致的性能提升。

## 全文

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.

---

引用前请核对原文;本文由星际流动整理:https://feed.astralor.com/items/itm_7e154ed693460246899c1dcb