摘要
[目的/意义]科学论文中的图、表等科学数据(文内数据)蕴含有丰富的知识内容。基于细粒度语义组织的文内数据发现有效聚合文献、科学数据两类科研产出,为未来实现全领域、全维度、全粒度、全类型的深度知识发现奠定基础。[方法/过程]梳理、对比、分析现有3类文内数据的发现模式,尝试构建面向细粒度内容描述的文内数据本体,揭示文内数据的显性特征、内容特征以及与其他类型科学产出的关联特征。[结果/结论]从面向应用的角度,提出基于本体的文内数据知识发现技术框架,细化“信息抽取—语义标注—关联应用”技术路线,针对关键点技术进行讨论。文内数据本体为科学论文内科学数据提供语义描述和关联组织依据,人工标注结合机器学习自动标注可以解决文内数据部分特征发现问题。
[Purpose/Meaning]The figures,tables and other scientific data in scientific papers(Scientific data in papers,SDIP)contain abundant knowledge.SDIP discovery based on fine-grained semantic organization can effectively aggregate literature and scientific data,laying a foundation for future in-depth knowledge discovery in all fields,all dimensions,all granularity and all types.[Method/Process]By combing,comparing and analyzing the existing organization and discovery patterns of three types of SDIP,this paper attempted to construct SDIP ontology that oriented to fine-grained knowledge description,and revealed SDIP s the explicit characteristics,content characteristics and characteristics correlation with other types of scientific output.[Result/Conclusions]From the perspective of application,this paper proposed an ontology-based framework for SDIP knowledge discovery,elaborated the technical route of“information extraction-semantic annotation-relational application”,and discussed the key technologies.Ontology provided semantic description and relevance discovery basis for SDIP.Manual annotation and automatic annotation using machine learning algorithm could solve the some problem of SDIP feature discovery.
作者
丁培
Ding Pei(National Science Library,Chinese Academy of Sciences,Beijing 100190,China;School of Economics and Management,University of Chinese Academy of Sciences,Beijing 100190,China;Library,Shenzhen University,Shenzhen 518060,China)
出处
《现代情报》
CSSCI
2020年第2期34-43,共10页
Journal of Modern Information
基金
广东省哲学社会科学规划学科共建项目“支持深度知识发现的文内数据与文献关联研究”(项目编号:GD18XTS07)
教育部人文社会科学研究青年基金项目“基于成熟度视角的高校图书馆科学数据管理服务能力评价研究”(项目编号:19YJC870028)
关键词
文内数据
科学数据
科学论文
本体组织
数据发现
scientific data in scientific papers
scientific data
scientific literature
ontology organization
data discovery
