摘要
针对古代科技文献知识组织中存在的数据多源异构和语义关联缺失等问题,提出一种基于异质信息网络的知识挖掘与可视化方法。首先设计领域知识表示模型并构建初始知识库;其次采集在线百科数据,通过规则模板与大语言模型从中抽取三元组;最后将三元组数据集转换为异质信息网络,对度分布、中心性和社区结构等关键指标进行分析。基于多维数据集构建了可视化应用,能直观呈现古代科技体系各知识单元的语义关联,为古代科技文献的数字化组织与知识发现提供工具支撑。
To address the issues of multi-source heterogeneity and semantic association deficiency in the knowledge organization of ancient scientific and technological documents,this paper proposes a knowledge mining and visualization approach based on heterogeneous information networks.Firstly,a domain knowledge representation model is designed to construct an initial knowledge base.Next,online encyclopedic data are collected,from which triples are extracted through rule templates and large language models.Finally,the triple dataset is transformed into a heterogeneous information network,and key metrics such as degree distribution,centrality,and community structure are analyzed.A visualization application is built based on multi-dimensional datasets to intuitively present the semantic relationships among knowledge units in the ancient scientific and technological system,providing tool support for the digital organization and knowledge discovery of ancient scientific and technological documents.
作者
潘俊
胡鹏飞
陶祥兴
Pan Jun;Hu Pengfei;Tao Xiangxing
出处
《新世纪图书馆》
2025年第8期70-78,共9页
New Century Library
基金
国家社会科学基金项目“古代科技文献名物知识图谱构建与人文计算研究”(项目编号:23BTQ019)的研究成果之一。
关键词
异质信息网络
关系抽取
古代科技文献
知识挖掘
数字人文
Heterogeneous information network
Relation extraction
Ancient Chinese scientific and technological documents
Knowledge mining
Digital humanities
