地理学报

地理学报 ›› 2017, Vol. 72 ›› Issue (4): 737-752.doi: 10.11821/dlxb201704014

• 地理信息 • 上一篇    下一篇

全球科研论文合作网络的结构异质性及其邻近性机理

刘承良1,2(), 桂钦昌1(), 段德忠1,2, 殷美元1   

  1. 1. 华东师范大学城市与区域科学学院,上海 200241
    2. 华东师范大学科技创新与发展战略研究中心,上海 200062
  • 收稿日期:2016-06-14 修回日期:2016-11-25 出版日期:2017-04-20 发布日期:2017-04-20
  • 作者简介:

    作者简介：刘承良(1979-), 男, 博士, 副教授, 硕士生导师, 中国地理学会会员(S110009837M), 主要从事经济地理复杂性研究。E-mail: clliu@re.ecnu.edu.cn

  • 基金资助:
    国家自然科学基金项目(41571123, 41471108)

Structural heterogeneity and proximity mechanism of global scientific collaboration network based on co-authored papers

Chengliang LIU1,2(), Qinchang GUI1(), Dezhong DUAN1,2, Meiyuan YIN1   

  1. 1. School of Urban and Regional Science, East China Normal University, Shanghai 200241, China
    2. Insititute for Innovation and Strategic Studies, East China Normal University, Shanghai 20062, China
  • Received:2016-06-14 Revised:2016-11-25 Published:2017-04-20 Online:2017-04-20
  • Supported by:
    National Natural Science Foundation of China, No.41571123, No.41471108

RichHTML

PDF (PC)

赞

被引次数

24 | 33

摘要：

以科研论文为媒介的知识合作网络已成为知识溢出的重要通道,但目前学术界对全球科研合作网络结构的复杂性涌现机制缺乏深入的探讨。基于2014年Web of Science核心合集所收录的科研论文合著数据,借助大数据挖掘技术、复杂网络、空间统计和重力模型分析,刻画了全球科研论文合作网络的拓扑结构、空间格局及其邻近性机理。结果发现：① 拓扑结构上,形成了以美国为核心的层级网络,具有小世界性和等级层次性,发育出典型的等级“核心—边缘”结构。② 空间格局上,以美国、西欧、中国和澳大利亚为顶点的“四边形”成为全球科研论文合作网络的骨架;三大中心性指标值的空间分异明显,强度中心性形成以美国为极核,加拿大、澳大利亚、中国及西欧诸国为次中心的“一超多强”格局,与之类似的介数中心性呈现北美、西欧和东亚“三足鼎立”的形态,度中心性分布则相对均匀,表现出“大分散、小集中”的“多中心—边缘集散”格局。③ 重力回归分析发现,地理距离抑制了国际科研论文合作,不过其影响力较弱;社会与经济邻近性对全球科研论文合作具有明显的促进作用,语言差异不是国际科研合作交流的障碍。

关键词: 科研论文合作网络, 结构异质性, 复杂网络分析, 重力模型, 邻近性机制

Abstract:

Despite increasing importance of academic papers in global knowledge flows, the structural disparities and proximity mechanism related to international scientific collaboration network attracted little attention. To fill this gap, based on data mining from Thomson Reuters' Web of Science database in 2014, its heterogeneities in topology and space were portrayed using visualizing tools such as Pajek, Gephi, VOSviewer, and ArcGIS. Topologically, 211 countries and 9928 ties are involved in global scientific collaboration network, but the international network of co-authored relations is mono-centricand dominated by the United States. It exhibits some features of a "small-world" network with the smaller average path length of 1.56 and the extremely large cluster coefficient of 0.73 compared to its counterpart, as well as the better-fitting exponential distribution accumulative nodal degree. In addition, the entire network presents a core-periphery structure with hierarchies, which is composed of 13 core countries and the periphery of 198 countries. Spatially, densely-tied and high-output areas are mainly distributed in four regions: West Europe, North America, East Asia and Australia. Moreover, the spatial heterogeneity is also observed in the distributions of three centralities. Amongst these, the countries with greater strength centrality are mainly concentrated in North America (i.e. the US and Canada), Western Europe (i.e. the UK, France, Germany, Italy and Spain), and China, noticeably in the US, which forms the polarizing pattern with one superpower of the US and great powers such as China and the UK. Similarly, the big three regions consisting of West Europe, North America and Asian-Pacific region have the peak betweenness centrality as well. Slightly different from the two above, the distribution of nodal degree centrality is uneven in the world, although regional agglomeration of high-degree countries is still observed. Last but not least, the proximity factors of its structural inequalities were also verified by correlational analysis, negative binomial regression approach and gravity model of STATA. The findings further confirm that geographical distance has weakened cross-country scientific collaboration. Meanwhile, socio-economic proximity has a positive impact on cross-country scientific collaboration, while language proximity plays a negative role.

Key words: scientific collaboration network, structural heterogeneity, complex network