文章摘要

夏翠娟.GenAI环境下多模态文化记忆资源知识表示研究:以古琴减字谱为例[J].中国图书馆学报,2026,52(2):126~154
GenAI环境下多模态文化记忆资源知识表示研究:以古琴减字谱为例
Knowledge Representation of Multimodal Cultural Memory Resources in the GenAI Environment:The Guqin Subtractive Notation as a Case Study
投稿时间：2025-08-13  修订日期：2025-10-29
DOI：
中文关键词: 匹配、概念匹配的检索和数据科学驱动的量化计算和可视化分析,到大模型驱动的相似性计算、跨模态检索和自然语言交互的转向。本研究形成的多元知识表示方法的领域应用解决方案和GenAI垂直领域应用路径具有一定的可扩展性和可移植性,对GenAI在数字人文领域应用有一定参考借鉴意义,也为未来研究提供了可供批判和演化发展的基础。图7。表6。参考文献28。关键词  数字人文  数字记忆  多模态文化记忆资源  知识表示  生成式人工智能  古琴减字谱
英文关键词: Digital humanitiesDigital memoriesMultimodal cultural memory resourcesKnowledge representationGenerative artificial intelligenceGuqin subtractive notation
基金项目:
作者单位
夏翠娟 中国人民大学信息资源管理学院 北京 100872 
摘要点击次数: 226
全文下载次数: 184
中文摘要:
      在现有数字人文研究与实践中,文本化、结构化、语义化知识表示方法已成为常规范式,但还需以跨越式发展方式突破现有困境,以适应GenAI时代数字人文研究与实践的新需求。本研究采用“数据建模—界面批判”实验研究法,以古琴减字谱这一典型的多模态文化记忆资源为例,构建综合运用语义化和向量化表示法的多元共生互补的知识表示体系框架,并将其应用于古琴减字谱识谱和释谱场景中；面向应用场景,构建了包含大规模样本的古琴减字谱多模态资源、结构化数据集、知识图谱和训练数据集,形成了“多模态大模型大样本微调训练+向量数据库存储检索+基于知识图谱的检索增强生成”的垂直领域应用解决方案,实现了从文本匹配、
英文摘要:
GenAI technology has been used to efficiently solve difficult problems in the field of natural sciences,and will have an undeniable impact on research in the humanities and social sciences,especially on the knowledge representation of multimodal cultural memory resources at the intersection of digital humanities. The existing textual,structured,and semantic knowledge representation methods that have become the conventional paradigm in digital humanities research and practice urgently require upgrading and replacement through leapfrog development to break through existing difficulties,while adapting to the new needs of digital humanities research and practice in the GenAI era,and even helping to address the shortcomings of GenAI technology.
This study mainly adopts the experimental research method of “data modeling-interface criticism”,which is suitable for digital humanities research,and adopts a spiral iterative optimization approach. Taking the typical multimodal cultural memory resource of Guqin subtractive notation as an example,this study constructs a multidimensional symbiotic and complementary knowledge representation system framework that comprehensively applies the semantic vectorization representation method. The constructed knowledge representation system was applied to the scenario of Guqin subtractive notation recognition and interpretation.
To address this application scenario,a multimodal resource repository,a structured dataset,a knowledge graph,and a training dataset of Guqin subtractive notation containing large scale samples were constructed,forming a vertical domain application solution of “multimodal large scale sample fine tuning training + vector database storage retrieval + knowledge graph guided retrieval augmented generation”,thereby exploring the vertical domain application path of GenAI technology. This study achieves a shift from text matching,keyword matching,concept matching retrieval,and data driven quantitative computing and visualization analysis to large model driven similarity calculation,cross modal retrieval,and natural language interaction. The application solution and technological application path formed in this study exhibit certain scalability and portability,providing valuable reference for the application of GenAI in vertical domains of digital humanities,and also providing a foundation for future critical reflection and evolutionary development. 7 figs. 6 tabs. 28 refs.
查看全文   查看/发表评论  下载PDF阅读器