摘要
[目的/意义]图书情报领域的知识计量研究长期停留在知识载体计量层面,难以深入挖掘细粒度的知识属性特征。文章试图从学术文献的文本片段层面开展知识单元粒度的知识计量研究。[方法/过程]首先,提出基于“人工标注—规则归纳—机器识别—规则补充”流程的知识单元抽取方法;其次,通过该方法构建知识描述规则库,并抽取知识单元;最后,对知识单元的主题、结构等属性信息进行数理统计分析,得到知识单元在这些属性上的分布特征。[结果/结论]基于“人工标注—规则归纳—机器识别—规则补充”流程构建的知识描述规则库收敛性较好,可以较为准确地从学术文献中提取知识单元文本片段。在此基础上进行的知识单元分布特征分析,能够较为准确地揭示文献知识单元在主题、结构上的分布特征,真正将知识计量的粒度从知识载体缩小至知识单元。
[Purpose/significance]The research of knowledge measurement in Library and Information Science always stays at measurement level of knowledge carriers,and it is difficult to recognize fine-grained characteristics of knowledge.This paper attempts to go deep into the text fragments of academic literature and carry out the quantitative measurement research of knowledge element granularity.[Method/process]Firstly,propose an extraction method of knowledge element based on the process of“manual annotation-rule induction-machine recognition-rule improvement”.Then,rule base for knowledge descriptions is constructed and knowledge element is extracted by this method.Finally,topic,structure and other attributes of knowledge elements are analyzed by mathematical statistics,and the distribution characteristics of knowledge elements on these attributes are obtained.[Result/conclusion]Experiment proves that the knowledge description rule base based on the process of“manual annotation-rule induction-machine recognition-rule improvement”has good convergence,and can extracts knowledge elements from academic literature accurately.Based on these extracting results,the distribution characteristics of knowledge elements in topic and literature structure are analyzed.In this paper,the granularity of knowledge measurement is really extended from knowledge carrier to knowledge element.
出处
《情报理论与实践》
CSSCI
北大核心
2023年第4期90-98,共9页
Information Studies:Theory & Application
基金
河南省哲学社会科学规划项目“基于词汇功能识别的交叉学科知识涌现机制研究”(项目编号:2022BZH008)
中央高校基本科研业务费资助(创新资助)项目“文献内容视角下的知识单元数量、关系、质量与价值计量研究”(项目编号:2022CXZZ108)
中央高校基本科研业务费项目“开源跨模态科技情报知识组织与智能分析”(项目编号:CCNU22QN016)的成果之一。
关键词
知识单元
知识抽取
知识计量
知识分布
学术评价
自然语言处理
knowledge element
knowledge extraction
knowledge measurement
knowledge distribution
academic evaluation
natural language processing
