| 陈涛,张欣,冯卓彤,杨鑫.文化遗产多模态资源知识统一表征模型构建研究[J].中国图书馆学报,2025,51(6):60~77 |
| 文化遗产多模态资源知识统一表征模型构建研究 |
| Construction of Knowledge Unified Representation Model for Multimodal Resources of Cultural Heritage |
| 投稿时间：2024-12-10 修订日期：2025-04-16 |
| DOI： |
| 中文关键词: 文化遗产 多模态资源 知识表征模型 数字人文 |
| 英文关键词: Cultural heritage Multimodal resources Knowledge representation model Digital humanities |
| 基金项目: |
|
| 摘要点击次数: 128 |
| 全文下载次数: 93 |
| 中文摘要: |
多模态是物理对象的真实写照和科学研究对象的常态。文本、图像、音频、视频和3D模型等多模态资源是中华文化全景呈现的关键,也是全面激活文化资源、深挖文化价值的重要所在。本文构建了文化遗产多模态资源知识统一表征模型（N ary）,首先设计N ary本体结构,包含集合类、资源类、模态类、形态类和注释类五大核心类。在此基础上,兼顾不同模态资源的特色,从知识组织角度对多模态资源内容进行统一表征。N ary表征模型在传统RDF描述框架的基础上,扩展出时间范围（δ）、图像区间（ω）、空间方位（α）和时序（τ）属性,旨在用统一的知识形式表示多模态知识内容。知识统一表征模型是构建高质量数据集的基础,也为跨模态知识交互提供了底层逻辑支撑。最后从文化遗产智慧保护、文化遗产要素内容表达、中华文化传播效能和生成式AI数据基石等角度探讨N ary表征模型在助推文化遗产强保护、高质量发展方面的潜在价值。图7。表3。参考文献21。
|
| 英文摘要: |
Cultural heritage is an important source for promoting Chinese cultural heritage and strengthening cultural confidence. Multimodal resources,such as text,images,audio,video and 3D models,serve as authentic representations and research standards for cultural heritage objects. They are not only important for fully activating cultural resources and deeply exploring cultural values,and key to the panoramic presentation of Chinese culture,but also represent the inevitable path toward the era of general artificial intelligence.
This study focuses on the cultural heritage resource modality and its internal knowledge positioning,representation,organization and integration,aiming to complement the modal structure of existing cultural heritage models. It first outlines the four layer structure of the knowledge representation of cultural heritage from a macro perspective:collection layer,resource layer,morphology layer and annotation layer,and designs the corresponding N ary ontology structure. The N ary ontology consists of five core classes:collection class,resource class,modal class,morphology class and annotation class. It should be emphasized that the N ary ontology model designed here is not a conventional knowledge representation model in the field of cultural heritage,nor is it mutually exclusive with the existing domain models. Instead,it is a cross modal knowledge integration framework. Its intersection with domain models such as CIDOC CRM and EDM occurs at the annotation layer,providing knowledge representation at the modal level for the knowledge description of existing domain models.
Based on the N ary ontology,this paper introduces the N ary knowledge unified representation model from the perspective of knowledge organization,designed for the unified representation of multimodal knowledge. The N ary unified representation model takes multimodal resource content as its starting point and extends the image interval （ω）,time range （δ）,spatial direction （α）,and time sequence （τ） markers on the basis of the RDF （Resource Description Framework） description of the content. Additionally,the paper proposes the analytical idea of n ary2rdf and introduces the concept of multimodal time series knowledge graph. Multimodal time series knowledge graph takes into account the dual characteristics of multimodal and time sequence,highlighting the evolution of cultural heritage from a dynamic and multi dimensional perspective.
Finally,the study discusses the potential value of N ary representation model in promoting the robust protection and high quality development of cultural heritage from the perspectives of smart cultural heritage protection,heritage element content expression,the efficiency of Chinese culture communication and the data foundation for Generative AI. Different from the existing domain models,the N ary unified knowledge representation model is a rediscovery of the “modalities” in the knowledge organization of cultural heritage. It is not only the basis of constructing high quality datasets,but also expected to provide underlying logical support for cross modal knowledge interaction. 7 figs. 3 tabs. 21 refs.
|
|
查看全文
查看/发表评论 下载PDF阅读器 |
|
|
|
京公网安备11010802047316号