arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2303.05063cs.CL
收藏

ICL-D3IE：用于文档信息抽取的、具备多样化示例更新能力的上下文学习

ICL-D3IE: In-Context Learning with Diverse Demonstrations Updating for Document Information Extraction

University of Electronic Science and Technology of China（电子科技大学）；Singapore Management University（新加坡管理大学）；Beijing Forestry University（北京林业大学）；Beijing Rongda Technology Co., Ltd.（北京融达科技有限公司）

Jiabang He, Lei Wang, Yi Hu, Ning Liu, Hui Liu, Xing Xu, Heng Tao Shen

备注：
ICCV 2023. Code is available at https://github.com/MAEHCM/ICL-D3IE
阅读 PDF
arXiv 页面 ↗HTML 原文 ↗

AI总结：

本文提出 ICL-D3IE 框架，通过困难、关系和格式化示例及迭代更新增强 LLM 的文档信息抽取能力，并在 ID/OOD 基准上超越全量微调方法。

摘要：

诸如 GPT-3 和 ChatGPT 之类的大型语言模型（LLM）已通过上下文学习在各种自然语言处理（NLP）任务上展现出显著成果；上下文学习是指基于少量示例进行推理。尽管它们在 NLP 任务中取得了成功，但尚无研究评估 LLM 利用上下文学习执行文档信息抽取（DIE）的能力。将 LLM 应用于 DIE 面临两个挑战：模态差距和任务差距。为此，我们提出了一个简单但有效的上下文学习框架，名为 ICL-D3IE，它使 LLM 能够借助不同类型的示例执行 DIE。具体而言，我们从困难训练文档中提取最困难且最具区分度的片段，作为有利于所有测试实例的困难示例。我们设计了描述关系的示例，使 LLM 能够理解位置关系。我们引入格式化示例，以便于答案抽取。此外，该框架通过迭代更新多样化示例来改进它们。我们在三个广泛使用的基准数据集上的实验表明，与以往使用完整训练进行微调的预训练方法相比，ICL-D3IE 框架使 Davinci-003/ChatGPT 在分布内（ID）设置和分布外（OOD）设置下均取得更优性能。代码可在 https://github.com/MAEHCM/ICL-D3IE 获取。

英文摘要：

Large language models (LLMs), such as GPT-3 and ChatGPT, have demonstrated remarkable results in various natural language processing (NLP) tasks with in-context learning, which involves inference based on a few demonstration examples. Despite their successes in NLP tasks, no investigation has been conducted to assess the ability of LLMs to perform document information extraction (DIE) using in-context learning. Applying LLMs to DIE poses two challenges: the modality and task gap. To this end, we propose a simple but effective in-context learning framework called ICL-D3IE, which enables LLMs to perform DIE with different types of demonstration examples. Specifically, we extract the most difficult and distinct segments from hard training documents as hard demonstrations for benefiting all test instances. We design demonstrations describing relationships that enable LLMs to understand positional relationships. We introduce formatting demonstrations for easy answer extraction. Additionally, the framework improves diverse demonstrations by updating them iteratively. Our experiments on three widely used benchmark datasets demonstrate that the ICL-D3IE framework enables Davinci-003/ChatGPT to achieve superior performance when compared to previous pre-trained methods fine-tuned with full training in both the in-distribution (ID) setting and in the out-of-distribution (OOD) setting. Code is available at https://github.com/MAEHCM/ICL-D3IE.

机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。

↑