跳转至

CLAIM：基于不确定性度量的大语言模型开放域主动澄清框架

文章背景与核心概要

在开放域人机交互中，大语言模型（LLM）常因用户查询的模糊性或不完整性而产生泛化过度或错误的回答。虽然通过主动提问进行澄清可以显著提升交互质量，但现有的方法往往高度依赖昂贵的人工标注和偏好对齐，这限制了其在实际场景中的泛化能力和部署效率。

本文提出了 CLAIM 框架，这是一种基于不确定性驱动的开放域主动澄清学习方法。CLAIM 的核心创新在于通过多模型回答差异的熵值来量化查询的不确定性，从而完全摆脱了对人工标注的依赖。该框架通过结合监督微调（SFT）和组相对策略优化（GRPO），构建了一个高效且稳健的澄清决策模型，为大模型在开放域交互中的主动理解提供了低成本的解决方案。

论文元数据 (Paper Metadata)

  • arXiv ID: 2608.11631
  • 主要学科: 人工智能 (cs.AI)
  • ACM 分类: I.2.7
  • 提交日期: 2026年8月12日
  • 作者: Kuangzhao Yang, Ziliang Zhao, Zhicheng Dou
  • arXiv ID: 2608.11631
  • Primary Subject: Artificial Intelligence (cs.AI)
  • ACM Classes: I.2.7
  • Submission Date: 12 August 2026
  • Authors: Kuangzhao Yang, Ziliang Zhao, Zhicheng Dou

摘要 (Abstract)

在开放域人机交互场景中，大语言模型（LLM）经常遇到模糊或不完整的用户查询。在这种情况下，直接生成答案往往会导致回答过于笼统、错误或信息量不足。相比之下，提出澄清性问题可以显著提高交互质量。

In open-domain human-computer interaction scenarios, large language models (LLMs) frequently encounter user queries that are ambiguous or incomplete. In such cases, directly producing an answer often leads to overgeneralized, erroneous, or low-information responses. In contrast, asking clarifying questions can substantially improve interaction quality.

然而，现有方法仍然严重依赖人工标注数据或偏好对齐来解决两个基本挑战：何时需要澄清，以及应该澄清查询的哪个方面。这种依赖导致了高昂的标注成本，并限制了模型的泛化能力。

However, existing approaches still rely heavily on manually annotated data or preference alignment to address two fundamental challenges: when clarification is necessary, and which aspect of the query should be clarified. This reliance incurs high annotation costs and limits generalization.

为了应对这些挑战，我们提出了 CLAIM，这是一个用于开放域主动澄清学习的不确定性驱动框架。CLAIM 通过量化多模型回答差异所产生的熵来衡量查询的不确定性，从而消除了对显式人工偏好标注的需求。这种不确定性信号随后被用于构建高质量的合成数据，通过监督学习和强化学习的结合，实现了一个统一澄清决策模型的训练。

To address these challenges, we propose CLAIM, an uncertainty-driven framework for active clarification learning in open-domain settings. CLAIM eliminates the need for explicit human preference annotations by quantifying query uncertainty through the entropy induced by answer disagreements across multiple models. This uncertainty signal is then used to construct high-quality synthetic data, enabling the training of a unified clarification decision model through a combination of supervised learning and reinforcement learning.

具体而言，我们提出了一种熵驱动的合成数据生成流水线，该流水线集成了基于熵的不确定性估计、语义聚类和基于推理的判断，实现了对澄清需求的可靠自动标注。为了训练 CLAIM，我们将澄清过程表述为一个结构化的决策生成问题，并采用了一种结合监督微调（SFT）和组相对策略优化（GRPO）的训练范式。

Specifically, we propose an entropy-driven synthetic data generation pipeline that integrates entropy-based uncertainty estimation with semantic clustering and reasoning-based judgments, enabling reliable automatic annotation of clarification requirements. To train CLAIM, we formulate the clarification process as a structured decision generation problem and adopt a training paradigm that combines supervised fine-tuning (SFT) with group-relative policy optimization (GRPO).

实验结果表明，CLAIM 能够在不依赖人工标注数据的情况下学习到稳定且具有泛化能力的澄清策略，为大语言模型在现实世界开放域交互中的主动理解提供了一种低成本且稳健的解决方案。

Experimental results demonstrate that CLAIM can learn stable and generalizable clarification strategies without relying on manually labeled data, offering a low-cost and robust solution for proactive understanding in real-world open-domain interactions with LLMs.

全文及访问链接 (Full-Text & Access Links)