图书情报工作 ›› 2020, Vol. 64 ›› Issue (8): 111-124.DOI: 10.13266/j.issn.0252-3116.2020.08.013

• 知识组织 • 上一篇    下一篇

基于语义和位置相似的作者共被引分析方法及效果实证

张汝昊   

  1. 中国科学院成都文献情报中心 成都 610041;中国科学院大学经济与管理学院图书情报与档案管理系 北京 100049;中国科学院文献情报中心 北京 100190
  • 收稿日期:2019-09-27 修回日期:2019-12-17 出版日期:2020-04-20 发布日期:2020-04-20
  • 作者简介:张汝昊(ORCID:0000-0002-4372-8726),硕士研究生,E-mail:zhangruhao@mail.las.ac.cn。

Empirical Study of a Semantic and Proximity-based Author Co-citation Analysis Method

Zhang Ruhao   

  1. Chengdu Library and Information Center, Chinese Academy of Sciences, Chengdu 610041 Department of Library and Information Science, School of Economics and Management, University of Chinese Academy of Sciences, Beijing 100049;National Science Library, Chinese Academy of Sciences, Beijing 100190
  • Received:2019-09-27 Revised:2019-12-17 Online:2020-04-20 Published:2020-04-20

摘要: [目的/意义] 作者共被引分析是探索领域知识结构的重要方法,在复杂的学科发展态势下,其依赖于共被引频次的作者关联度度量颇具争议。对此,提出一种基于语义和位置相似的作者共被引分析改良方法。[方法/过程] 在介绍基本原理的基础上,以图情领域为例开展基于语义和位置相似的作者共被引分析改良方法的效果实证,面向CNKI期刊库进行引文全文挖掘,并对引用句及引用位置进行抽取,结合预训练的领域词嵌入模型计算共被引文献间的深层相似度和作者间的关联强度,利用网络分析和因子分析法对比该方法与传统方法的效果差异。[结果/结论] 结果证明,基于语义和位置相似的作者共被引分析改良方法能更准确地识别共被引作者的关联强度,可发现更为细致的学科知识结构,并具有可拓展性与可应用性。

关键词: 作者共被引分析, 引文内容分析, 共引位置分析, 全文本引文分析, 领域知识结构

Abstract: [Purpose/significance] The author co-citation analysis is an vital method to explore the domain knowledge structure. In the context of complex development of disciplines, the author’s relevance measure based on the co-citation frequency is quite controversial. The study proposed an improved method for author co-citation analysis based on the similarity of content semantics and the proximity of locations. [Method/process] Based on the introduction of its basic principles, the field of LIS was used as an example to demonstrate the effect of the method, a full-text mining of citations for CNKI Chinese journals was conducted, and the citing sentences and reference positions were then extracted. Combined with pre-trained domain word embedding models, the deep correlation between the co-cited literature and the strength of the connection between the authors were measured. A network analysis and a factor analysis were then used to compare the differences on effects between the method and the traditional method. [Result/conclusion] The results show that the method can more accurately identify the correlation strength between authors, and find more detailed subject knowledge structure, and has a certain scalability and applicability.

Key words: author co-citation analysis, citation content analysis, co-citation proximity analysis, citation in full-text, knowledge structure

中图分类号: