用上下文和常识验证，精准识别大模型在企业场景中的幻觉内容。
HalluciNot: Hallucination Detection Through Context and Common Knowledge Verification
- 基于上下文和常识构建幻觉分类体系，细分四类幻觉类型。
- 提出新模型HDM-2，在多个数据集上优于现有方法，支持词级标注。
- 专为企业部署设计，兼顾效率、领域适配与错误精确定位。
本文提出一个面向企业场景的大语言模型幻觉检测系统。针对企业应用中的幻觉问题，构建了包含上下文依赖、常识性、企业特异性及无害性陈述的全新分类体系。所提出的幻觉检测模型HDM-2通过结合上下文与普遍事实验证，输出幻觉评分与词级标注，实现精准定位问题内容。为评估其在上下文与常识幻觉上的表现，我们构建了新数据集HDMBench。实验结果表明，HDM-2在RagTruth、TruthfulQA和HDMBench三个数据集上均优于现有方法。本工作还解决了企业部署中的计算效率、领域适应性和细粒度错误识别等挑战。相关评估数据集、模型权重与推理代码均已公开。
原文摘要 · Abstract (English)
This paper introduces a comprehensive system for detecting hallucinations in large language model (LLM) outputs in enterprise settings. We present a novel taxonomy of LLM responses specific to hallucination in enterprise applications, categorizing them into context-based, common knowledge, enterprise-specific, and innocuous statements. Our hallucination detection model HDM-2 validates LLM responses with respect to both context and generally known facts (common knowledge). It provides both hallucination scores and word-level annotations, enabling precise identification of problematic content. To evaluate it on context-based and common-knowledge hallucinations, we introduce a new dataset HDMBench. Experimental results demonstrate that HDM-2 out-performs existing approaches across RagTruth, TruthfulQA, and HDMBench datasets. This work addresses the specific challenges of enterprise deployment, including computational efficiency, domain specialization, and fine-grained error identification. Our evaluation dataset, model weights, and inference code are publicly available.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。