作者混合共引网络对知识图谱绘制的改进研究

发布时间：2018-06-01 19:06

  本文选题：作者共引网络 + 共引网络　； 参考：《图书情报工作》2017年03期

【摘要】：[目的/意义]作者共引网络分析(ACNA)是文献计量学中的重要分析方法,旨在通过寻找学术文献集合中作者之间的共引关系绘制出特定领域的知识图谱,进而指导科学研究。然而,ACNA的一个缺陷是其原始矩阵输入信息量过小。本文通过提出作者混合共引网络(HACNA),绘制更为精确的科学知识图谱。[方法/过程]鉴于不同种类的学术网络能为绘制知识图谱提供不同维度的信息,提高知识图谱绘制的精确性,本文以合著网络和引用网络为例,结合其他种类的学术网络在ACNA基础上进行精确科学知识图谱的绘制。[结果/结论]实证研究结果显示,与ACNA相比,HACNA绘制出的知识图谱在聚类过程中能够使得同类作者更为聚拢、不同类作者更为分散,从而提高了聚类效果和可视化程度。同时,HACNA绘制出的知识图谱还能够挖掘出更多细节。
[Abstract]:[objective / significance] the author cocitation network analysis (ACNA) is an important analytical method in bibliometrics, which aims to draw a map of knowledge in a specific field by looking for cocitation relationships among authors in a collection of academic documents, and then to guide scientific research. However, one of the defects of ACNA is that its original matrix input information is too small. A more accurate map of scientific knowledge is drawn through the author's hybrid cocitation network HACNA. [methods / processes] given that different types of academic networks can provide different dimensions of information for mapping knowledge maps and improve the accuracy of mapping knowledge maps, this paper takes coauthor networks and citation networks as examples. Combined with other types of academic networks to map accurate scientific knowledge on the basis of ACNA. [results / conclusion] the results of empirical study show that compared with ACNA, the knowledge map drawn by ACNA can make the same authors gather more closely, and the authors of different classes become more dispersed, thus improving the clustering effect and visualization degree. At the same time HACNA map of knowledge can also be mined out more details.
【作者单位】： 北京大学信息管理系;美国印第安纳大学信息学与计算机学院;
【基金】：中国科技信息研究所系所合作项目研究成果之一
【分类号】：G353.1
，

本文编号：1965299