摘要
h型中心性是近年来情报学中关于网络分析独特方法的有趣进展。本文使用“微机电系统（MEMS）”领域2001～2010年10年的数据集构建合作演化网络，通过Spearman秩相关验证了h型中心性在预测优先连接的有效性，并和度中心性进行了对比。结果显示，微机电系统领域数据集中的作者数和联系数符合优先连接的增长机制，度中心性和h型中心性方法都可以有效地预测优先连接。h型中心性方法可为预测优先连接提供更多的视角。其中，主要考虑加权网络节点联系的权重的一类方法（即h-Degree，a—Degree和g-Degree）预测结果与度中心性存在相当区别。而只考虑节点的相邻度的一类方法（即L—index，al-index和gl—index）和结合了节点的邻节点的度以及节点和邻节点之间的联系的权重的一类方法（即Hw-Degree，Aw-Degree和Gw-Degree）表现相对接近于度中心性，这两类方法可以作为合作网络演化发展趋势的预测的补充或替代参数。
In recent years, h-type centrality has become an interesting topic in unique network analysis methods in information science. We use MEMS field data sets from 2001 to 2010 to build scientific collaboration network evolution. We first test the mechanism of growth and then test the effectiveness of h-type centrality in predicting preferential attachment using the Spearman rank correlation in order to compare them with degree centrality. It was found that the number of authors and links accord with the mechanism of growth and h-type centrality approaches and degree centrality are efficient to predict the preferential attachment. The h-type centrality methods can provide more perspectives for the prediction of preferential attachment. The category of approaches （i. e. , h-Degree, a-Degree, g-Degree） is considerably different from degree centrality. Andthe category of approaches （i. e. , L-index, M-index, gl-index） to consider a node＇s neighbors＇ degree and the category of approaches （ i. e. , Hw-Degree, Aw-Degree, Gw-Degree） are closer to the degree centrality, so they can be taken as complementary or alternative parameters to predict accurately for the evolution of cooperation networks.
出处
《情报学报》
CSSCI
北大核心
2015年第2期156-163,共8页
Journal of the China Society for Scientific and Technical Information
基金
中央高校基本科研业务费No.CDJKXB12004资助
关键词
优先连接
h型中心性度
中心性网络
分析微机电系统
preferential attachment, h-type centrality, degree centrality, network analysis, MEMS
