新书推荐|《文化遗产领域的人工智能与智慧数据:全球进展与中国方案》
2026-07-02 14:20 湖北
《文化遗产领域的人工智能与智慧数据:全球进展与中国方案》作为《数字人文与智能计算》丛书的第二卷,由泰勒-弗朗西斯出版集团旗下的Routledge出版社以开放获取形式出版。
一、书籍简介
《文化遗产领域的人工智能与智慧数据:全球进展与中国方案》(AI and Smart Data for Cultural Heritage: Global Achievements and China’s Innovations)作为《数字人文与智能计算》丛书的第二卷,由泰勒-弗朗西斯(Taylor & Francis)出版集团旗下的Routledge出版社以开放获取形式出版。本书探究人工智能与智慧数据为数字人文(DH)研究带来的范式转型,汇集亚洲、欧洲、北美及南美多地的研究成果,以全球视野展现文化遗产领域的创新研究方法与实践路径。
从大数据演进至智慧数据、迈入人工智能时代的发展进程中,数字人文研究者持续开拓全新的研究概念、方法范式与实践形态。智慧数据可依托任意规模的可信、情境化、认知型、预测性且可直接应用的数据提炼研究洞见;在人工智能的协同赋能下,智慧数据将持续在本领域发挥核心价值。本书兼顾理论路径与实践应用,系统呈现数字人文领域中人工智能与智慧数据赋能文化遗产的前沿进展与发展趋势,同时也为文化遗产场景下运用此类技术的批判理论、方法体系与实践探索做出了重要学术贡献。
对于深耕技术与人文研究交叉领域的数字人文研究者、文化遗产专业从业者及人工智能技术实践者而言,本书是不可或缺的核心参考读物。
二、编者简介
王晓光,武汉大学信息管理学院院长、教授,同时担任武汉大学文化遗产智能计算实验室主任与数字人文研究中心主任,研究方向涵盖数字资产管理、知识组织、语义出版及数字人文。
曾蕾,美国肯特州立大学信息科学系教授,获美国匹兹堡大学博士学位。曾担任国际图联(IFLA)标引与分类法委员会主席、国际知识组织学会(ISKO)理事、国际信息科学与技术学会 (ASIS&T) 理事、都柏林核心元数据组织(DCMI)顾问委员会主席及执行委员会委员、W3C 图书馆关联数据(LLD)工作组特邀专家成员、国际联盟 iSchools 数字人文课程委员会(iDHCC)主席等。其研究领域包括知识组织系统、元数据、语义技术与数字人文,累计出版专著6部,发表学术论文百余篇。
高瑾,伦敦大学学院(UCL)数字人文研究中心(UCLDH)联合主任,伦敦大学学院信息研究系数字档案方向讲师,兼任维多利亚与阿尔伯特博物馆研究员。研究聚焦数字人文史、网络分析、数字化、来源研究与数据标准领域。
赵珂,武汉大学信息管理学院、文化遗产智能计算实验室助理研究员。她拥有武汉大学信息资源管理博士学位、伦敦大学学院数字人文理学硕士学位,以及分获中国与韩国院校的艺术学学士、工学学士学位,研究方向聚焦数字叙事、数字人文与人机交互。
三、出版信息
书名:《文化遗产领域的人工智能与智慧数据:全球进展与中国方案》(AI and Smart Data for Cultural Heritage: Global Achievements and China’s Innovations)
主编:王晓光,曾蕾,高瑾,赵珂
电子书出版时间:2026年5月25日
出版地:伦敦
出版社:Routledge
总页数:396页
电子书ISBN:9781003666530
开放获取资助方:武汉大学
四、书籍目录
第一部分 全球创新路径
第1章
人文学科的开放数据应用——来自《开放人文数据期刊》的观察
作者:Barbara McGillivray, Daniele Borkowski, Andrea Farina, Simon Mahony, Xiaoguang Wang
第2章
智能体人工智能在社科与人文数据发现中的应用
作者:Vassilis Routsis, Yujian Gan, Sagar Uprety, Antonis Bikakis
第3章
基于Arches平台的文化遗产资源管理——面向资源清单与概念型叙词表的智慧数据方案
作者:Dennis Wuthrich, Philip Carlisle, Kevin Kochanski
第4章
协同网络环境下的智慧数据:建筑与城市空间图像获取
作者:Artur Simões Rozestraten, Giselle Beiguelman, Vânia Mara Alves Lima
第5章
作为数据基础设施的馆藏资源——德国与澳大利亚的实践视角
作者:Marco Humbel, Julianne Nyhan, Nina Pearlman
第6章
泰国社区档案与文化遗产数字化技术
作者:Pimphot Seelakate
第7章
利用非完美AI成果深化佛教典籍研究——以SAT文本数据库为例
作者:Kiyonori Nagasaki
第8章
文化遗产项目开发中的伦理设计考量
作者:Lala Hajibayova
第二部分 中国创新实践
第9章
从故纸遗存到活态内核——人工智能时代文献遗产的智能化发展研究
作者:Li Niu, Chi Jin, Anrunze Li, Rundong Hu
第10章
面向中国古琴减字谱释读的知识增强多模态大语言模型
作者:Cuijuan Xia
第11章
基于大语言模型的非物质文化遗产文本自动词性标注
作者:Zhixiao Zhao, Dongbo Wang
第12章
数字化再现与活化——面向中国青铜器图像智能生成的扩散模型微调研究
作者:Xilong Hou, Xiaoguang Wang
第13章
中国古代科技文献数据库建设与知识挖掘
作者:Xiang Zheng, Mingjie Li
第14章
智慧元数据视角下价值增值驱动的北京传统村落本体建模
作者:Chunqiu Li, Jie Guo, Jiayi Wang, Wirapong Chansanam, Chen Chen
第15章
故宫博物院馆藏数据的重构与增益
作者:Yipei Ye
五、章节简介
第1章
人文学科的开放数据应用——来自《开放人文数据期刊》的观察
Open Data Adoption Across the Humanities:Insights from the Journal of Open Humanities Data
数字人文研究规模持续拓展,数据驱动研究方法与开放研究理念已逐步成为领域共识。开放数据与智慧数据是这一转型的核心支撑,与英国、欧盟及全球层面的研究数据开放出版行动方向一致。作为人文学科领域规模最大的跨学科数据期刊,《开放人文数据期刊》(JOHD)过去六年间发展迅速,折射出学界对开放数据的关注度持续上升。期刊内容展现了不同学科的多元实践路径,但开放数据的采纳程度与管理模式仍存在显著差异,也为数字人文社群带来了值得深入探讨的重要问题。
本章通过对JOHD刊载成果中的开放数据实践开展量化分析,探究人文学科各领域采纳、共享与出版开放数据的现状。研究发现,不同学科在数据集类型、规模与生成方式上存在明显的不均衡性,地理分布层面也存在代表性缺口。研究结论明确了领域未来发展的关键方向,对数字人文领域的开放数据建设具有广泛启示:尤其揭示了人工智能与智慧数据应用的不均衡现状,阐明了人工-自动化混合工作流在应对标注复杂度问题上的核心作用,同时指出部分学科与数据类型存在严重的代表性不足。
Digital Humanities (DH) research has steadily expanded, embracing data-driven methods and open research principles. Open and smart data play a key role in this shift, aligning with UK, European, and global efforts to publish research data openly. The Journal of Open Humanities Data (JOHD), the largest interdisciplinary data journal in the humanities, has grown rapidly over the past six years, reflecting a rising interest in open data. Its content reflects the range of approaches across disciplines, though adoption and management of open data still vary widely, raising important questions for the DH communities. This chapter explores how various humanities disciplines adopt, share, and publish open data with a quantitative analysis of open data practices in JOHD publications. Findings show significant disciplinary asymmetries in dataset types, sizes, and generation methods. We also identify gaps in geographical representation. These findings highlight critical areas for future development and have broader implications for open data in DH, particularly by revealing the uneven adoption of AI and smart data, the essential role of hybrid manual-automated workflows in managing annotation complexity, and the significant underrepresentation of certain disciplines and data types.
第2章
智能体人工智能在社科与人文数据发现中的应用
Agentic AI for Data Discovery in the Social Sciences and Humanities
大语言模型的最新进展推动了多智能体系统的发展,使其能够处理愈发复杂的领域专属任务。本章介绍CORDIAL-AI项目——这是一项结合检索增强技术、以API为核心驱动的研究项目,可通过自然语言交互访问高度复杂、粒度精细的英国人口流动普查数据。系统通过分层智能体解析用户意图、检索相关元数据并生成可执行查询,能够适配繁杂的代码表、层级嵌套的地理区划与多维度变量。研究重点关注可解释性、数据来源追溯与元数据整合三大维度。
通过低秩适应(LoRA)进行微调的实验表明,只要配备充足的合成数据与专家标注训练语料,小型开源大语言模型的数据检索表现可接近专有模型水平。初步实证基准测试显示,模型在变量选择、地理实体识别与API调用准确率上均有提升。本章将研究结论置于数字研究基础设施的更广泛方法论讨论中,强调批判性系统设计、严谨的数据治理与以用户为中心的评估,是推动数据密集型研究负责任发展的核心。CORDIAL-AI这类系统降低了复杂数据集的使用门槛,为社会科学与人文领域的研究开辟了新的可能。在注重语境、阐释与透明度的学科领域,此类工具能够帮助研究者提出更具深度的问题,更清晰地追溯研究结论的生成过程。
Recent advances in large language models have enabled the development of multi-agent systems capable of handling increasingly sophisticated and domain-specific tasks. This chapter explores CORDIAL-AI, a retrieval-augmented, API-driven project that provides natural language access to highly complex and granular UK census flow data. Layered agents parse user intent, retrieve relevant metadata, and generate executable queries that navigate extensive code lists, hierarchical and nested geographies, and multi-dimensional variables. Particular attention is given to explainability, provenance, and metadata integration. Fine-tuning via low-rank adaptation demonstrates that small open-source LLMs can approach the data retrieval performance of proprietary models when provided with sufficient synthetic and expert-annotated training corpora. Preliminary empirical benchmarks show improvements in variable selection, geographic recognition, and API-call accuracy. These findings are situated within broader methodological discussions on digital research infrastructures, emphasising the importance of critical system design, rigorous data curation, and user-centred evaluation in advancing responsible practices in data-intensive research. Systems like CORDIAL-AI make it easier to work with complex datasets, lowering technical barriers and opening new possibilities for research across the social sciences and humanities. In fields where context, interpretation, and transparency matter, such tools can help researchers ask more nuanced questions and more clearly trace how findings are produced.
第3章
基于Arches平台的文化遗产资源管理——面向资源清单与概念型叙词表的智慧数据方案
Managing Cultural Heritage Resources with the Arches Platform:Smart Data for Resource Inventories and Concept-based Thesauri
即便依托现代技术,对实物遗存、考古遗址等文化遗产资源进行记录登记依然是一项难题。文化遗产对象与场所的语境化描述复杂度高,遗产资源阐释本身存在固有不确定性,这都对主流数据管理平台的能力提出了挑战。本章阐述 Arches平台(www.archesproject.org)如何通过数据架构、知识表示标准、信息管理工作流与系统集成能力,应对文化遗产信息管理的诸多挑战,为可扩展、高性能的智慧数据系统建设提供支撑。
Arches采用CIDOC-CRM等本体构建文化遗产资源的语义数据模型,通过定义信息类与属性,搭建具备语境化、认知性与预测性的数据结构。平台的数据管理、发现、传播与集成能力,能够为学者与遗产资源管理者提供可信、相关且可直接应用的信息。Arches采用模块化、可扩展的架构设计,本章将重点介绍其在遗产资源清单、叙词表与受控词表等权威数据管理场景下的具体应用。
Documenting cultural heritage (CH) resources such as physical objects and archaeological sites, even with modern technologies, remains a challenge. The complexities of describing CH objects and places in context and the inherent uncertainties that come with interpreting heritage resources stretch the capabilities of most modern data management platforms. This chapter describes how the Arches platform (https://www.w3.org/1999/xlink" xlink:href="https://www.archesproject.org">www.archesproject.org) implements data schema, knowledge representation standards, information management workflows, and systems integration capabilities that meet the challenges of managing CH information and support the creation of scalable and fast smart data systems. Arches implements ontologies, such as the CIDOC-CRM, to create semantic data models of cultural heritage resources. These data models define the information classes and properties needed to structure contextualised, cognitive and predictive data. The platform’s data management, discovery, dissemination, and integration capabilities deliver the trusted, relevant, and consumable information demanded by both scholars and CH resource stewards. Arches is a modular and extensible platform, and this chapter will highlight specific Arches data management applications for heritage inventories and authority data such as thesauri and controlled vocabularies.
第4章
协同网络环境下的智慧数据:建筑与城市空间图像获取
Smart Data in Collaborative Web Environments to Access Architectural and Urban Spaces Images
十五年来,ARQUIGRAFIA项目持续建设智慧数据资源,致力于巴西建筑与城市空间文化遗产的保存与开放获取。项目对实体图像馆藏进行数字化处理,同时开放用户自由上传数字图像,在此基础上搭建了一整套摄影信息表示与检索工具体系。为引导用户细致观察图像、对照片中的建筑与城市空间形成自主判断,ARQUIGRAFIA构建了一套基于造型-空间二元属性的建筑主观阐释内部体系。与此同时,项目还研发了知识组织系统,明确了描述图像的术语定义与术语间关联关系。
项目也持续孵化各类试点项目,例如将音频讲解与地理标记摄影图像相关联的“露天博物馆”项目。目前,项目团队与数字馆藏研究组合作开展人工智能技术实验,测试多种计算机视觉模型对ARQUIGRAFIA馆藏图像的分析效果。这些实验提取了艺术与设计维度的相关数据,挖掘出原始编目体系中未能呈现的馆藏解读层次。
For fifteen years, the ARQUIGRAFIA project has been creating smart data to preserve and enable access to the cultural heritage of Brazilian architecture and urban spaces. This path provided a set of tools for representing and retrieving photographic information, as a result of the digitisation of physical collections of images as well as digital images freely uploaded by users. The ARQUIGRAFIA has thus developed an internal system of subjective interpretations of architectures based on binomials of plastic-spatial qualities in order to encourage users to closely observe images and formulate judgments about buildings and urban spaces represented in photographs. Additionally, a Knowledge Organisation System has been developed, considering the definitions of terms and their interrelationships that describe such images. Synchronously, it has functioned as an incubator for pilot projects such as Open Air Museum, that associates audio messages with georeferenced photographic images. Currently, experiments with AI resources in partnership with the Digital Collections and Research team have tested different Computer Vision models to analyse the set of images available in the ARQUIGRAFIA. These experiments have collected data on art and design, uncovering layers of the collection’s legibility that were not evident in its original cataloguing.
第5章
作为数据基础设施的馆藏资源——德国与澳大利亚的实践视角
Collections as Data Infrastructures;Perspectives from Germany and Australia
参与国家及国际数字基础设施项目的文化遗产机构(图书馆、档案馆、博物馆、高校馆藏与植物标本馆等)有着怎样的实践经验?哪些因素推动或制约了文化遗产机构打通分散的馆藏数据孤岛?不同国家的影响因素存在何种差异?
2020年代初,全球掀起“馆藏即数据”(Collections as Data)基础设施的建设热潮,例如英国的“迈向国家馆藏”计划、欧盟的“欧洲文化遗产共同数据空间”项目,以及澳大利亚研究数据共享平台等。然而,尽管文化遗产机构是数字基础设施的核心数据供给方,目前仍缺乏跨国比较的定性研究来呈现机构的实践视角。厘清这一视角,有助于更深入地理解机构推进馆藏数据框架建设时面临的潜在且关键的挑战。
本章首次就馆藏即数据基础设施建设的国际经验展开对话研究,在英国相关调研结论的基础上,补充德国与澳大利亚两国文化遗产机构的两组焦点小组访谈数据。研究结论涵盖基础设施项目的顶层设计路径、学科文化差异、职业发展体系不足与伦理问题等多个维度。
What are the experiences of cultural heritage organisations (libraries, archives, museums, university collections and herbaria) participating in national and international digital infrastructure projects? Which factors enable and impede cultural heritage organisations in unifying siloed collections? How do these factors differ between countries? The early 2020s were marked by a surge of investments in Collections as Data infrastructures, such as made by the United Kingdom’s Towards a National Collection programme, the European Commissions’ Common European Data Space for Cultural Heritage, or the Australian Research Data Commons. However, despite cultural heritage organisations’ key role as data providers for digital infrastructures there is a lack of qualitative research comparing their perspectives internationally. Ascertaining such a perspective is necessary to better understand the latent and yet potent challenges that are encountered by institutions as they seek to advance collections data frameworks. This chapter takes the first steps in bringing international perspectives on Collections as Data infrastructure developments in conversation. Findings from UK consultations are advanced with data gathered in two focus group discussions held with cultural heritage organisations in Germany and Australia. The findings address issues in the way infrastructure programmes tend to be set up, divergent disciplinary cultures, deficiencies in professional development and ethics.
第6章
泰国社区档案与文化遗产数字化技术
Community Archives in Thailand and Digital Technology for Cultural Heritage
本章梳理了泰国社区档案的发展简史,并提出界定泰国社会档案特征的概念框架。尽管社区档案在泰国的整体存在感较弱,但部分社区档案已成为泰国社会中最具创新性的档案实践形态。作者通过文献研究,梳理了泰国社区档案与文化遗产馆藏建设中技术应用的相关案例,揭示了数字时代泰国社区档案运用数字技术开展档案工作的动因与实践路径。
This chapter presents a short history of community archives in Thailand and a conceptual framework for defining the characteristics of archives in Thai society. Although it seems that community archives are under-represented in Thailand, some community archives are found to be among the most innovative archives in society. The author uses a literature review to explore some Thai case studies on the usage of technologies in the provisioning and analysis of community archives and cultural heritage collections. The findings reveal the motives and methods of Thai community archives to use digital technology for their archival practices in the Digital Age.
第7章
利用非完美AI成果深化佛教典籍研究——以SAT文本数据库为例
Enhancing Buddhist Scripture Research with Imperfect AI Outcomes:A Case Study of the SAT Text Database
本章探讨如何将高精度AI光学字符识别(OCR)技术与国际图像互操作框架(IIIF)相结合,部署以文献为核心的检索增强生成(RAG)系统,借助“存在瑕疵但具备实用价值”的AI输出,切实提升佛教典籍研究效率。
研究以SAT大藏经文本数据库与社群编目平台为基础,将日本国立国会图书馆的AI-OCR技术应用于大型木刻大藏经图像数据集,实现了OCR识别文本与SAT带行号基准文本的逐段对齐。文本到IIIF图像的点击跳转功能,解决了长篇典籍中段落定位的“最后一公里”难题——即便OCR结果仍有残余误差,也可作为高效索引指向原始文献证据。本章公布了不同版本OCR的准确率基准,介绍了逐行渐进修正OCR结果的编辑工作流,并探讨了符合文本编码倡议(TEI)规范的异文记录方案。
此外,研究还推出了“Bauddha AI”RAG服务,可整合通过Apache Solr检索到的领域文献结论,生成带文献来源与链接支撑的内容摘要。作者认为,将存在瑕疵的OCR结果与权威文本对齐、以精选文献约束生成内容,二者结合能够构建可靠、可供学者核验的研究工作流。尽管该方法仍存在覆盖范围有限、多语种扩展不足、大语言模型表现不稳定等局限,但可迁移至其他人文学科领域,降低文献溯源与文献综述的成本。
This chapter examines how “imperfect yet practical” AI outputs can measurably enhance Buddhist scripture research by integrating high-accuracy AI-OCR with IIIF and by deploying a literature-centred RAG system. Building on the SAT Daizōkyō Text Database and a community curation platform, this study applies the National Diet Library’s AI-OCR to large woodblock Tripiṭaka image sets and implements passage-level alignment between OCR text and SAT’s line-numbered main text. Click-through navigation from text to IIIF images closes the “last-mile” gap of locating passages within long volumes, turning OCR – despite residual errors – into an efficient pointer to material evidence. The paper reports accuracy benchmarks across OCR versions, describes an editing pipeline that incrementally corrects OCR at the line level, and discusses TEI-conformant strategies for recording variants. Complementarily, it introduces “Bauddha AI”, a RAG service that synthesises findings from domain papers retrieved via Apache Solr, enabling sourced, link-backed summaries. The author argues that (1) aligning imperfect OCR with trusted texts, and (2) constraining generation with curated literature, jointly produce reliable, scholar-auditable workflows. While limitations remain – coverage, multilingual expansion, evolving LLM behaviour – the approach is portable across humanities domains and lowers the cost of source verification and literature synthesis.
第8章
文化遗产项目开发中的伦理设计考量
Consideration of Ethical Design in the Development of Cultural Heritage Initiatives
本章提出,知识表示、组织与发现系统的设计必须纳入伦理考量,因为这类系统同时兼具技术属性与社会属性。基于其社会技术属性,本章重点探讨这类系统如何体现、支撑与约束价值敏感设计(Value Sensitive Design)的核心价值,包括个人隐私尊重、所有权与财产权、无偏见原则、普遍可用性、信任机制、知情同意与环境可持续性。在人工智能技术飞速发展的当下,要推动这些价值落地,需要通过实证研究与理论建构,全面评估价值敏感设计在弥合社会价值与技术进步分歧方面的潜力。
In this chapter it is argued that ethical considerations in the design of knowledge representation, organisation, and discovery systems are essential, as these systems encompass both technical and social dimensions. Given their socio-technical nature, the focus of the discussion is an examination of how these systems embody, support, and constrain core values central to Value Sensitive Design, including respect for individual privacy, ownership and property rights, freedom from bias, universal usability, trust, informed consent, and environmental sustainability. To promote these values in this era of rapidly advancing AI, empirical inquiry and theoretical development are needed for thorough assessment of the potential of Value Sensitive Design to mediate potential divergences between social values and technological progress.
第9章
从故纸遗存到活态内核——人工智能时代文献遗产的智能化发展研究
From Dusty Pages to Living Essence:A Study on the Intelligent Development of Documentary Heritage in the Era of AI
本章探讨人工智能时代文献遗产的智能化转型路径,阐释其在保存与活化集体文化记忆中的核心作用。研究以文化计算与语义技术相关理论为基础,构建了包含模式、能力、资源三个维度的三维发展模型。该模型强调文化基因提取、知识逻辑组织与情境化用户服务三大核心。
本章借助大语言模型等前沿工具,回应遗产获取、语义理解与公众参与层面的核心挑战。研究以入选《世界记忆名录》的苏州丝绸档案为案例,展示了如何对历史文献进行语义结构化处理、嵌入动态知识网络,并通过智能化服务面向多元受众传递价值。研究表明,人工智能能够弥合认知鸿沟、提升阐释能力,推动文献遗产的可持续发展与创造性转化。本章最终提出,未来的文化记忆不应仅停留在保存层面,更应借助数字智能实现主动活化,让人们能够与历史建立更深入、更具价值的联结。
This chapter explores the intelligent transformation of documentary heritage in the era of artificial intelligence, highlighting its role in preserving and revitalising collective cultural memory. Drawing on theories of cultural computing and semantic technologies, it constructs a three-dimensional development model comprising the model, capability, and resource dimensions. The model emphasises the extraction of cultural genes, logical organisation of knowledge, and contextualised user services. Through advanced tools such as LLMs (Large Language Models), the chapter addresses key challenges in heritage access, semantic understanding, and public engagement. Using the Suzhou Silk Archives in the Memory of the World as a case study, it demonstrates how historical documents can be semantically structured, embedded in dynamic knowledge networks, and delivered through intelligent services to diverse audiences. The study illustrates the potential of AI (Artificial Intelligence) to bridge cognitive gaps, enhance interpretability, and promote the sustainable development and creative transformation of documentary heritage. Ultimately, this chapter advocates for a future in which cultural memory is not merely preserved but actively revitalised through digital intelligence, enabling deeper, more meaningful engagement with the past.
第10章
面向中国古琴减字谱释读的知识增强多模态大语言模型
A Knowledge-enhanced Multi-modal Large Language Model for Chinese Guqin Subtractive Notation Interpretation
古琴是中国最古老的传统乐器之一,音色雅致、文化象征意蕴深厚,是典型的多模态文化记忆资源。多模态大语言模型的发展,为多模态文化记忆资源的知识服务提供了新的解决方案。然而,对于古琴减字谱相关的文本、图像、音视频等特殊文化记忆资源,现有多模态大语言模型的知识表示能力仍需优化,才能达到预期效果。
本研究以多模态古琴减字谱资源为训练数据,结合古琴减字谱知识图谱,探索多模态大语言模型在文化遗产领域的垂直应用路径。研究的最终目标是借助生成式人工智能技术,通过知识增强的多模态大语言模型,帮助更多人理解古琴减字谱。本研究搭建的释读应用场景表明,大语言模型在解决判别类问题时表现优异;但在自然语言交互的生成类问题中,结合知识图谱技术与人类专家构建的形式化知识,能够显著提升大语言模型在垂直任务中的准确率、可靠性与专业性。
Guqin is one of the oldest traditional Chinese musical instruments, known for its refined sound and deep cultural symbolism as a typical multimodal cultural memory resource. The development of multimodal large language models (MLLMs) provides new solutions for the knowledge service of multimodal cultural memory resources. However, for the knowledge representation of some special cultural memory resources such as text, images, audio, and video resources related to the Guqin Subtractive Notation, the existing MLLMs need further optimisation to achieve the expected results. This study uses multi-modal Guqin Subtractive Notation resources as training data, and combines them with the Knowledge Graph of Guqin Subtractive Notation to explore a vertical application path of MLLMs in the field of cultural heritage. The ultimate goal of this study is to use the knowledge-enhanced MLLMs to help more people understand the Guqin Subtractive Notation with the help of Generative AI technologies. The practical construction of the interpretation application scenario in this study demonstrates that LLMs perform better in solving discriminative problems, but in addressing generative issues in natural language interaction, combining Knowledge Graph technology and the formalised knowledge generated by human experts can significantly enhance the accuracy, reliability, and professionalism of LLMs in vertical tasks.
第11章
基于大语言模型的非物质文化遗产文本自动词性标注
Automatic Part-of-Speech Tagging of Intangible Cultural Heritage based on Large Language Models
大语言模型引发了自然语言处理领域的技术变革,也推动了各垂直领域应用的发展。本研究探索将大语言模型应用于非物质文化遗产领域的自动文本标注——该领域拥有专业词汇体系与独特文化语境。
研究分为三个阶段:第一,从中国非物质文化遗产官方网站采集非遗文本数据,通过人机协作方式构建高质量标注数据集;第二,针对非遗文本设计零样本、单样本、少样本等不同提示格式的微调数据;第三,采用低秩适应(LoRA)与全参数微调两种方式,测试Qwen与GLM模型在不同训练数据量下的表现。与低秩适应相比,全参数微调对提示一致性的要求更高,鲁棒性更弱。研究结论为优化大语言模型的精准文本标注能力提供了参考,对非物质文化遗产及其他专业领域均具有借鉴意义。本研究凸显了微调策略、数据准备与指令结构设计对提升模型表现的重要价值。
Large language models (LLMs) have revolutionised natural language processing, enabling advancements in domain-specific applications. This study explores the use of LLMs for automatic lexical annotation in intangible cultural heritage (ICH), a field with specialised vocabulary and cultural context. The research follows three phases: (1) Collecting ICH data from China’s official Intangible Cultural Heritage website and creating a robust annotation dataset through human-machine collaboration. (2) Developing fine-tuning data with various prompting formats (0-shot, 1-shot, 3-shot) for ICH texts. (3) Evaluating Qwen and GLM models using LoRA and full-parameter fine-tuning to assess performance across training data volumes. Compared to LoRA, full parameter fine-tuning has higher requirements for prompt consistency and poorer robustness. These results provide insights into optimising LLMs for precise lexical annotation, with implications for ICH and other specialised fields. The study highlights the importance of fine-tuning strategies, data preparation, and instruction structuring in enhancing model performance.
第12章
数字化再现与活化——面向中国青铜器图像智能生成的扩散模型微调研究
Digital Recreation and Revitalisation:Fine-tuning Diffusion Models for the Intelligent Generation of Chinese Bronze Vessel Images
生成式人工智能技术为文化遗产的保护、修复与传播带来了新机遇。本章以中国青铜器为研究案例,探究生成式人工智能在图像智能生成领域的变革潜力,及其在文化遗产领域的创新应用路径。
研究构建了中国青铜器领域本体,实现青铜器图像的多维度语义表示;提出一种大语言模型辅助的人机协同语义标注方法,提升语义标注效率。在此基础上,研究利用低秩适应算法对Stable Diffusion模型进行微调,实现了高质量、风格统一的中国青铜器图像生成。微调后的模型在生成内容的视觉质量与图文匹配度上均有显著提升,为文物保护、修复与文化传播提供了新的工具与技术路径。最后,本章探讨了人工智能驱动图像智能计算的底层逻辑与实现路径,阐明其在推动文化遗产数字化再现与活化中的作用。生成式人工智能不仅深化了对文化文物的语义理解与计算分析,也推动了文化遗产智慧数据资源的建设,进而促进文化遗产的创新复用与活化发展。
Generative AI technologies bring new opportunities for the preservation, restoration, and dissemination of cultural heritage. Focusing on Chinese bronze vessels as a case study, this chapter investigates the transformative potential of generative AI in intelligent image generation and its innovative application within cultural heritage domain. A domain ontology for Chinese bronze vessels is constructed to enable muti-dimensional semantic representation of bronze vessel images. A human-AI collaborative semantic annotation method, enhanced by large language models, is proposed to improve the efficiency of semantic annotation. Furthermore, by leveraging the LoRA algorithm to fine-tune the Stable Diffusion model, this study achieves high-quality and style-consistent image generation of Chinese bronze vessels. The fine-tuned model significantly enhances both the visual quality and text-image alignment of generated content, offering new tools and technical pathways for artifact preservation, restoration, and cultural communication. Finally, this chapter discussed the underlying logic and implementation pathways of AI-driven intelligent image computing, and highlighting their role in facilitating the digital recreation and revitalisation of cultural heritage. Generative AI not only strengthens the semantic understanding and computational analysis of cultural artifacts, but also facilitates the development of smart data resources for cultural heritage, thereby advancing the innovative reuse and revitalisation of cultural heritage.
第13章
中国古代科技文献数据库建设与知识挖掘
Database Construction and Knowledge Mining of Ancient Chinese Scientific and Technological Documents
为保存中国古代科技文献、盘活历史资源、支撑中国科技史研究,本章建设并研究了中国古代科技文献数据库(ACSTDD)。数据库收录了1912年以前(不含1912年)天文、地理、农业、水利、数学、物理、化学、生物、医药等多个科技领域的古代文献,提供分类浏览、组合检索等功能,同时配备辅助学术研究的相关工具。
基于该数据库,本章开展古代科技文献书目提要的知识挖掘实践,以清代蚕桑、棉花类农书为例,实现了作者、内容、地理分布等维度的知识重组与可视化呈现。本章弥补了中国古代科技文献组织程度不足的研究空白。通过数据库对分散的历史资源进行系统化整理,突破了古代科技文献碎片化分布导致无法开展大规模比较研究的局限,为科技史跨学科研究开辟了新路径。
To preserve ancient Chinese scientific and technological documents, utilise historical resources, and support research in the history of Chinese science and technology, this chapter constructs and studies the Ancient Chinese Scientific and Technological Documents Database (ACSTDD). The database contains ancient documents from various scientific and technological fields, including astronomy, geography, agriculture, hydraulics, mathematics, physics, chemistry, biology, and medicine before 1912 ce (excluding 1912 ce). It provides features, such as classification browsing and combination search. The system also provides tools to assist academic research. Based on ACSTDD, this chapter carries out the knowledge mining practice of bibliographic summaries of ancient scientific and technological documents. This chapter uses sericulture, cotton, and mulberry agricultural documents in the Qing dynasty as examples to achieve knowledge reorganisation and visual presentation related to authors, content, and geographical distribution. This chapter fills the gap and addresses the issue of the insufficient organisation of ancient Chinese scientific and technological documents. The systematic organisation of these scattered historical resources through ACSTDD enables large-scale comparative studies that were previously impossible due to the fragmented nature of ancient Chinese scientific and technological documents, opening new avenues for interdisciplinary research in the history of science and technology.
第14章
智慧元数据视角下价值增值驱动的北京传统村落本体建模
Value Addition-driven Ontology Modelling of Beijing Traditional Villages from the Perspective of Smart Metadata
传统村落蕴含古建筑、民俗等丰富的乡村文化资源,凭借其历史、文化与艺术价值备受学界与社会关注。本章阐释智慧元数据的核心内涵,提出价值增值驱动的北京传统村落本体建模方案。从智慧元数据视角出发,对传统村落的海量历史文化资源进行系统化组织,能够推动“数据—信息—知识—智慧”的层级转化,实现文化的代际传承与价值提升。
本章从官方平台“中国传统村落数字博物馆”采集北京市国家级传统村落的相关数据,从多格式资源中提取事件、民俗、故事、文化发展、历史沿革等知识实体粒度的信息。所构建的本体定义了村落信息、文化遗产、社会关系、历史发展、地理空间五大核心类,用于规范描述村落相关维度的内容。在智慧元数据的理论指引下,通过打通传统村落各类资源的关联链路,清晰呈现村落文化资源间的内在互动关系。本章研究为传统村落文化知识发现与文化的历时性传承提供了参考思路。
Traditional villages are rich in rural culture resources, such as ancient architecture and folklore, and have attracted attention due to their historical, cultural and artistic value. This chapter explains smart metadata and proposes value addition driven ontology modelling for traditional villages in Beijing. From the perspective of smart metadata, organising the abundant historical and cultural resources in a traditional village facilitates the transformation of “data-information-knowledge-wisdom”, enabling intergenerational cultural inheritance and value enhancement. This chapter collects data on national-level traditional villages in Beijing from the official website named “Traditional Chinese Village Digital Museum”. Events, folklore, story, cultural development, history, and other information at knowledge entity granularity will be extracted from the collected resources in various formats. The proposed ontology in this chapter defines core classes to describe village information, culture heritage, social relation, historical development, and geography. With the guidance of smart metadata and the interconnections of traditional village resources, interactive relationships between village cultural resources are expressed. This chapter offers insights for the discovery of traditional village cultural knowledge and cultural inheritance over time.
第15章
故宫博物院馆藏数据的重构与增益
Reconstruction and Enhancement of the Palace Museum's Collection Data
本文以故宫博物院为研究案例,回应数字时代博物馆面临的信息管理挑战。研究在CIDOC CRM框架下,构建了适配中国古代文物的专属本体模型——中国古代文物概念参考模型。通过编制受控词表、引入知识图谱技术,将馆藏编目数据重构为标准化、结构化的知识图谱。
这一转型优化了故宫博物院馆藏线上平台的检索服务,提升了检索效率与双语检索能力。此外,结构化数据可与大语言模型深度融合,提升机器理解、逻辑推理与内容生成效果。故宫博物院的实践实现了博物馆知识的创新组织与馆藏数据的价值挖掘,将传统编目数据转化为可计算的三元组结构,适配人工智能时代的数据处理需求。本研究不仅提升了故宫博物院的信息管理效率,也为其他文化遗产机构提供了可借鉴的实践经验。
This paper, taking the Palace Museum as a case, addresses museums’ information-management challenges in the digital age. It explores the development of an ontology model, Ancient Chinese Artifacts Conceptual Reference Model, tailored for ancient Chinese artifacts within the CIDOC CRM framework. Controlled vocabularies are compiled, and knowledge graph technology is introduced to restructure collection cataloguing data into a standardised and structured graph. This transformation enables enhanced search services on the Palace Museum’s collection online platform, including improved retrieval and bilingual search capabilities. Moreover, it facilitates the integration with large language models for better computer understanding, logical reasoning, and content generation. The Palace Museum’s practice innovatively organises museum knowledge and extracts value from collection data, converting traditional cataloguing data into a computable triple structure, thus meeting the AI-era’s data-processing demands. This research not only boosts the Palace Museum’s information-management efficiency but also provides valuable insights for other cultural heritage institutions.
六、学者评价
武汉大学副校长陆伟在本书的新书发布会视频致辞中向伦敦国王学院给予本次发布会的支持表示感谢。他表示,该书的出版是武汉大学信息管理学院和文化遗产智能计算实验室在国际数字人文领域研究的重要成果。未来,武汉大学将持续推进跨学科创新与国际学术合作,为全球文化遗产的数智化转型与人类文明的可持续发展贡献出中国智慧与力量。
Graham Wynn在致辞中表示,KCL高度重视国际学术交流,此次新书发布会是两校深化合作的重要契机,期待未来双方在教学、科研和人才培养上持续拓展合作空间。
Carolyn Kirby表示,该丛书的持续出版彰显了数字人文领域的蓬勃活力,泰勒-弗朗西斯(Taylor & Francis)出版集团将继续支持这一开放学术成果的发表,推动全球范围内的知识共享与学术对话。
信息管理学院院长、文化遗产智能计算实验室主任、丛书主编王晓光教授表示,人工智能技术的快速演进正在深刻重塑数字人文的研究范式与方法路径。文化遗产智能计算实验室紧抓学术前沿机遇,组织编写并出版此书,汇聚四大洲40位学者的学术贡献,标志着实验室在国际学术协作网络建设上的进一步发展。未来实验室将继续深化跨国、跨学科的合作机制,推动丛书向着更具前瞻性和包容性的方向发展,为数智赋能文化发展提供中国方案。
以上内容来自书籍官网
经数字人文资讯小编翻译整理而成
如需转载,请后台私信联系
编辑丨洪冰凤
校对丨罗斯鹏
排版丨魏翔