| 刘炜,金家琴,张宁,王丽华.从分类主题到语义空间:人工智能驱动的图书馆知识组织体系转型与发展[J].中国图书馆学报,2026,52(4):40~62 |
| 从分类主题到语义空间:人工智能驱动的图书馆知识组织体系转型与发展 |
| From Classification Schemes to Semantic Spaces:The Transfor mational Logic and Development Path of Library Knowledge Organization Systems in the AI Environment |
| 投稿时间：2026-02-24 修订日期：2026-03-13 |
| DOI： |
| 中文关键词: 图书馆 知识组织 人工智能 语义空间 大语言模型 知识标注 人机协作 关联数据 |
| 英文关键词: Library Knowledge organization Artificial intelligence Semantic spaces Large language models Knowledge annotation Human machine collaboration Linked data |
| 基金项目: |
|
| 摘要点击次数: 193 |
| 全文下载次数: 126 |
| 中文摘要: |
生成式人工智能的快速发展正在从底层逻辑上重塑图书馆知识组织的技术基础与价值定位。通过文献综述与多案例比较分析方法,系统梳理近年来国内外知识组织领域的理论进展与技术实践,从知识表示与处理方式的演进视角出发,分析符号操作、统计学习与语义计算等不同方法路径在知识组织中的作用与变化。研究发现:第一,传统知识组织的根本困境并非只是效率问题,而是符号体系与知识本体之间存在不可消除的语义鸿沟,表现为标引成本的指数增长、受控词表的语义僵化以及人工标注的系统性偏差；第二,AI技术并未使知识组织变得不必要,而是推动其核心功能从“文献揭示”转向“知识标注”,从服务人类检索转向同时服务机器理解；第三,传统知识组织资产对AI的支撑价值在当前数据稀缺阶段具有不可替代性,但随着合成数据与自监督学习的成熟,该价值将逐步递减。图书馆知识组织正经历从符号逻辑向向量语义的范式转换,图书馆界应在制度设计、人机回环技术工作流以及领域知识工程师能力重塑三个维度推进战略转型,适应未来人工智能环境中的方向性变革。图2。表2。参考文献50。
|
| 英文摘要: |
The rapid development of generative artificial intelligence（AI） is reshaping the technical foundation and value positioning of library knowledge organization from the bottom up. This paper aims to clarify the structural dilemmas faced by traditional knowledge organization systems in the AI environment,assess the substantial impact of AI technologies on the objectives of knowledge organization,and explore the internal logic and feasible path for transitioning from the “classification scheme” paradigm to the “semantic space” paradigmThrough literature review and comparative case analysis,this paper systematically examines recent theoretical developments and technical practices in the field of knowledge organization,comparing nine representative AI application cases worldwide,involving institutions including OCLC,the National Library of Finland,the Library of Congress,the National Library of China and so on. Based on the global library survey data from Clarivate and domestic empirical studies,the paper examines the role and changes of different methodological approaches in knowledge organization from the perspective of knowledge representation and processing,including symbolic operations,statistical learning,and semantic computing.
The paper reaches the following three conclusions:First,the fundamental dilemma of traditional knowledge organization is not just an issue of efficiency,but an irreducible semantic gap between the symbolic system and the knowledge ontology,manifested in the exponential increase in indexing costs,the semantic rigidity of controlled vocabularies,and the systematic biases in manual annotations. Second,AI technologies have not made knowledge organization unnecessary but have shifted its core function from “document representation” to “knowledge annotation,” moving from serving human retrieval to simultaneously serving machine understanding. Third,traditional knowledge organization assets （classification systems,authority files,controlled vocabularies） provide indispensable support to AI in the current data scarce stage. However,as synthetic data and self supervised learning mature,this value will gradually decline. Library knowledge organization is undergoing a paradigm shift from symbolic logic to vector semantics. In this process,while knowledge organization will not disappear,its functional positioning,technical methods,and target users will transition from ordered arrangements designed for human readers to semantic annotation and knowledge enhancement aimed at human machine hybrid ecosystems. The library sector should advance strategic transformation in three dimensions:institutional design,human in the loop （HITL） technical workflows,and the reskilling of domain knowledge engineers,in order to adapt to the directional changes in the AI environment of the future. 2 figs. 2 tabs. 50 refs.
|
|
查看全文
查看/发表评论 下载PDF阅读器 |
|
|
|
京公网安备11010802047316号