多重关系下的机构网络学科显著性研究

Research on Domain Visibility of Institution Networks with Multiple Relations

在学术机构的聚类研究中,研究人员发现单一角度的网络关系并不能充分揭示网络中个体间的相互关系,而在现实世界中同一个研究集合中的样本之间的多重关系是普遍存在的。为了评估不同的关系网络对机构之间的学科结构特征进行揭示的有效性,本研究构建了以学科一致性指标为代表的大学网络学科显著性模型,并从传统文献和Web文献两种不同途径,对342家机构之间因引用、合著和链接等不同关系生成的关系网络进行了对比。研究结果发现,链接网络在揭示机构之间存在的学科相关性时要优于合著网络和引用网络,而合著网络比较高的地域一致性说明合著行为很大程度上受到地域因素的影响。

In research into the clustering of academic institutions, researchers have found that it is insufficient to recognize patterns clearly based on the number of nodes connected through a single relation. In the real world, the phenomenon of multiple connections among a set of entities is a ubiquitous occurrence. To assess the capability of identifying sub-structures from different networks, a model measured based on the domain consistency as a means to demonstrate the visibility of a research domain is proposed in this study. Moreover, three networks at 342 institutions were built based on citations, co-authorships, and hyperlinked relations extracted from publications and web pages, and their identified sub-structures were compared. The results show that a hyper-link network is the best when compared to a co-author network or citation network for discovering hidden domain-similarity based patterns among academic institutions, whereas the co-authoring motivations presented in a co-author network among institutions are significantly affected by geographical factors.