---
id: 20260510-T0-04
title: "BALAR：用贝叶斯主动推理让LLM在多轮对话中高效追问，减少盲目猜测"
title_en: "BALAR: Bayesian Reasoning Helps LLMs Ask Better Questions in Dialogues"
url: https://ai.daily.yangsir.net/daily/20260510-T0-04
issue_date: 2026-05-10
publish_date: 2026-05-09T04:00:00.000Z
category: research
source_name: "arXiv cs.AI"
source_url: https://arxiv.org/abs/2605.05386
---

# BALAR：用贝叶斯主动推理让LLM在多轮对话中高效追问，减少盲目猜测

现有大语言模型在多轮交互任务中多采用被动响应，缺乏系统性推理机制来决定何时以及如何向用户获取信息，常出现盲目猜测。BALAR构建了一种基于贝叶斯的智能体主动推理循环，让模型能根据当前不确定性的变化进行动态提问。该方法显著降低了模型在信息不完整时产生幻觉的概率，对开发客服机器人或交互式诊断工具具有直接的实用价值。

## English Version

**BALAR: Bayesian Reasoning Helps LLMs Ask Better Questions in Dialogues**

Current LLMs in interactive settings often react passively and lack a principled mechanism to determine when or how to ask users for clarification, leading to blind guesses. BALAR introduces a Bayesian agentic loop for active reasoning, enabling models to dynamically ask questions based on evolving uncertainty. This approach reduces hallucinations under incomplete information, offering high practical value for customer service bots and diagnostic tools.

---

**来源**：[arXiv cs.AI](https://arxiv.org/abs/2605.05386)

**详情页**：https://ai.daily.yangsir.net/daily/20260510-T0-04

---

*智语观潮 · Daily — https://ai.daily.yangsir.net/llms.txt*