区间分割概念格及其在作者共被引分析中的应用  

The Interval Partition Concept Lattice and Its Application in Author Co-citation Analysis

作　　者：周奇志[1] 李秀霞[2] 程结晶[2] 

机构地区：[1]曲阜师范大学图书馆,日照276826 [2]曲阜师范大学传媒学院,日照276826

出　　处：《情报杂志》2016年第9期146-151,157,共7页Journal of Intelligence

基　　金：国家社会科学基金项目"文献内容分析与引文分析融合的知识挖掘与发现研究"(编号:16BTQ074)的阶段性成果之一

摘　　要：[目的/意义]作者共被引(Author Co-citation Analysis,简称ACA)是一种揭示作者关系和研究领域知识结构的重要方法,为丰富ACA方法的研究手段,将概念格引入ACA。基于概念格不能对多值属性数据集聚类的局限性,提出区间分割概念格,并给出定量计算ACA中各研究力量的计算公式。[方法/过程]以学科服务为研究主题,首先选取核心作者,然后提取核心作者相似矩阵的因子成份5个,之后对5个连续的因子属性值按照一定的阈值分别进行区间分割,利用Con Exp-1.3构建区间分割概念格,最后截取各因子属性对应的作者类团,同时定量给出不同作者类团对应研究方向的研究力量。[结果/结论]与传统的聚类环境SPSS、Ucinet6.0、Matlab相比,区间分割概念格不仅能够发现研究主题各研究方向的核心作者,还能发现各研究方向的领军人物,得到各研究方向量化后的研究力量。区间分割概念格可以对多值属性问题聚类,聚类结果可靠,可视化效果好,该方法还可以推广应用到文献分析、共词分析、期刊分析的聚类研究中。[ Purpose/Significance] The author co-citation analysis, viz ACA, is an important method to reveal the relationship between the author and the knowledge structure of the research domain. In order to enrich the research methods of ACA, the concept lattice is intro- duced into ACA in this paper. As the limitation of the concept lattice, the interval partition concept lattice is proposed. And the calculating formula of the research strength in ACA is given. [ Method/Process] Taking discipline service as the research subject, firstly, we select the core author, and extract five factor components of the core of the similarity matrix. Then the interval partition for six successive factor attribute values is cut apart respectively. Finally, we construct the interval partition concept lattice and intercept the author group of each factor attribute in ConExp- 1.3. Lastly, we give out the research strength of the research direction. [ Result/Conclusion ] Compared with the traditional clustering environment such as SPSS, Matlab and Ucinet6.0, the interval partition concept lattice can not only find the core author of the research topics, but also find the leading figure in the research direction, and can obtain the research strength of the various research directions. The interval partition concept lattice can deal with the problem of multi valued attributes, clustering results are reliable,and the method can be applied to the clustering process of literature analysis, co-word analysis and periodical analysis.

关 键 词：因子分析 区间分割 概念格 作者共被引分析 聚类分析 

分 类 号：G251.1[文化科学—图书馆学]