情报科学 ›› 2021, Vol. 39 ›› Issue (7): 30-37.

• 理论研究 • 上一篇    下一篇

学术交叉领域的双向聚类及主题关联特征分析

  

  • 出版日期:2021-07-16 发布日期:2021-07-16

  • Online:2021-07-16 Published:2021-07-16

摘要: 【目的/意义】为了发现交叉学科领域学术群及前沿研究主题,更有效地揭示交叉领域研究的核心研究者,
进一步发现核心学者与前沿主题之间的关联特征,促进国家重大交叉科研项目的合作攻坚,发现潜在的合作前景
并有效组织前沿重大交叉合作团队。【方法/过程】就地理信息领域内的关键词、作者、文献、期刊以及机构等信息进
行分析,采用双聚类分析的方法,实施了作者与主题的双向聚类分析。以此为基础研究了高产作者与研究主题的
关联特征,揭示出作者与主题间的关联关系。【结果/结论】结果表明,双向聚类方法具有突出的研究效果,既发现了
领域核心学术群体,又发现了前沿主题。进一步的关联分析发现核心学者群与前沿主题之间的关联特征,从而可
挖掘出隐性的跨学科主题与学者之间的关联。【创新/局限】从横向(研究者)与纵向(研究主题)两个方面进行双向
聚类分析,并同时考察两方面的关联是本文的创新之处。

Abstract: 【Purpose/significance】In order to discover interdisciplinary academic groups and frontier research topics, more effectively
reveal the core researchers of interdisciplinary researchers, further discover the correlation characteristics between core scholars and
frontier topics, promote the cooperation of national major interdisciplinary research projects, discover potential cooperation prospects,
and effectively organize frontier major interdisciplinary cooperation teams.【Method/process】This paper analyzes the key words, au?
thors, literatures, journals and institutions in the field of geographic information, and adopts the method of double cluster analysis to
implement the bi-directional cluster analysis of authors and topics. On this basis, the paper studies the characteristics of the relation?
ship between high-yield authors and research topics, and reveals the relationship between authors and research topic.【Result/conclu?
sion】The results show that the bidirectional clustering method has outstanding research effect, which not only finds the core academic
groups in the field, but also finds the frontier topics. Further association analysis finds the association characteristics between core
scholars and frontier topics, thus mining the implicit association between interdisciplinary topics and scholars.【Innovation/limitation】
The innovation of this paper is to conduct two-way cluster analysis from the horizontal (researcher) and vertical (research topic) as?
pects, and examine the correlation between the two aspects at the same time.