JieZi: A Large-Scale Expert-Audited Dataset and Benchmark for Ancient Chinese Character Exegesis

📄 arXiv: 2608.11741v1 📥 PDF

作者: Ran Li, Huiguo He, Jiahuan Cao, Junle Liu, Hiuyi Cheng, Lianwen Jin

分类: cs.CV, cs.AI

发布日期: 2026-08-12

备注: 19 pages, 13 figures. Accepted to the Dataset Track of ACM Multimedia 2026 for oral presentation

🔗 代码/项目: GITHUB


💡 一句话要点

提出JieZi数据集以解决古代汉字释义研究的不足问题

🎯 匹配领域: 支柱九:具身大模型 (Embodied Foundation Models)

关键词: 古代汉字 释义研究 多模态学习 数据集构建 专家审核

📋 核心要点

  1. 现有的计算方法在古代汉字释义研究中存在局限,缺乏系统化的数据集和基准,无法支持全面的学术分析。
  2. 本文提出了古代汉字释义(ACCE)任务,并构建了JieZi-Dataset和JieZi-Bench,以支持多层次的释义过程。
  3. 实验结果显示,当前模型在基本任务上表现良好,但在复杂的字形分析和语义推理上仍需改进,微调后性能显著提升。

📝 摘要(中文)

古代汉字的学术释义需要整合视觉观察、语言分析和历史背景。然而,现有的计算方法主要集中在字符识别和检索等子任务上,缺乏全面学术分析所需的结构化数据集和基准。为了解决这一局限性,本文提出了古代汉字释义(ACCE)任务,并构建了JieZi-Dataset和JieZi-Bench两个资源,前者是首个大规模专家审核的VQA训练数据集,后者是与释义过程对齐的评估基准。实验表明,当前模型在基本识别上表现良好,但在字形分析、语义推理和历时理解上存在困难,通过在JieZi-Dataset上进行微调显著提升了各个层面的性能。

🔬 方法详解

问题定义:本文旨在解决古代汉字释义研究中缺乏结构化数据集和基准的问题。现有方法主要集中于字符识别,无法满足学术分析的需求。

核心思路:通过构建古代汉字释义(ACCE)任务,整合视觉和语言信息,设计出多层次的释义过程,以提升对古代汉字的理解和分析能力。

技术框架:整体架构包括JieZi-Dataset和JieZi-Bench两个主要模块。JieZi-Dataset包含超过50万对问答对,经过专家审核和模板约束生成;JieZi-Bench则提供与释义过程对齐的评估标准。

关键创新:最重要的创新点在于构建了首个大规模、专家审核的VQA训练数据集,确保了数据的学术准确性和可靠性。与现有方法相比,提供了更系统化的释义分析框架。

关键设计:在数据集构建中,采用了专家设计的模板和源文本参考,减少了事实错误;每个关键阶段都进行了人工验证,以确保学术准确性。

🖼️ 关键图片

fig_0
fig_1
fig_2

📊 实验亮点

实验结果表明,当前的多模态大语言模型在基本识别任务上表现良好,但在字形分析、语义推理和历时理解方面存在困难。通过在JieZi-Dataset上进行微调,模型在所有四个层面的性能显著提升,展示了数据集的有效性和重要性。

🎯 应用场景

该研究的潜在应用领域包括古代汉字的学术研究、教育以及文化遗产保护等。通过提供系统化的释义工具,能够帮助研究人员更好地理解和分析古代文献,促进相关领域的发展。

📄 摘要(原文)

The scholarly exegesis of ancient Chinese characters demands integrating visual observation, linguistic analysis, and historical context. However, existing computational approaches focus narrowly on subtasks such as character recognition and retrieval, lacking the structured datasets and benchmarks required for comprehensive scholarly analysis. To address this limitation, we introduce Ancient Chinese Character Exegesis (ACCE), a vision-language question answering (VQA) task that models the scholarly exegesis process. ACCE is organized into four progressive levels: basic character identification, glyph-form analysis, meaning exegesis, and diachronic evolution analysis. To support this task, we construct two complementary resources. JieZi-Dataset is the first large-scale, expert-audited VQA training dataset for ACCE, comprising over 500K QA pairs. It is constructed via a pipeline that reduces factual errors by constraining generation with expert-designed templates and source-text references. Human verification is further applied at each key stage to ensure scholarly accuracy. JieZi-Bench is an evaluation benchmark aligned with the exegesis process, constructed and verified by human experts to ensure evaluation reliability. It consists of four levels with reference answers curated from authoritative lexicographic works held separate from the training data. Experiments on multimodal large language models show that current models perform well on basic identification but struggle with glyph analysis, semantic reasoning, and diachronic understanding. Fine-tuning on JieZi-Dataset substantially improves performance across all four levels. Code and dataset are available at https://github.com/Ran00w/JieZi.