情报科学 ›› 2023, Vol. 41 ›› Issue (2): 20-28.

• 理论研究 • 上一篇    下一篇

高校导师合作网络演化及形成机制研究 ——基于指数随机图模型

  

  • 出版日期:2023-02-01 发布日期:2023-04-07

  • Online:2023-02-01 Published:2023-04-07

摘要: 【目的/意义】合作是当今科学研究的重要方式,现有研究忽略了学者在研究生培养上的学术合作这一重要
形式。明确高校导师在研究生联合培养上的合作形成机制及演化特征对于深入理解科研合作及导师合作指导行
为具有重要意义。【方法/过程】以统计学科合作规模最大的三所高校为例,基于学位论文数据构建导师合作有向网
络。利用社会网络分析方法从宏观和微观层次探索网络演化规律和结构特征,基于指数随机图模型从网络结构和
导师个体属性角度明确合作的形成机制。【结果/结论】三所高校的导师合作网络具有相似的演化趋势,结构上呈现
集群性、链状和多个小团体的不同特征。结构效应对网络形成的影响在高校间存在差异,互惠性和中介性促进了
合作的形成,而闭合性的影响不显著。导师研究方向不具有同配倾向,职称等属性对网络形成具有显著的正向影
响。【创新/局限】扩展了学者合作研究范畴,结果对于发现导师合作规律和明确合作形成机制具有重要参考价值。
结论对于其他学科以及导师跨机构跨学科合作等特殊情形的适用性还有待考察。

Abstract: 【Purpose/significance】Collaboration is an important way of scientific research at present, and the existing research on sci?
entific collaboration ignores the important form of academic collaboration in joint postgraduate training. It is of great significance to
clarify the formation mechanism and evolution characteristics of collaboration between supervisors in joint postgraduate training for a
deep understanding of scientific collaboration and cooperative behavior of supervisors.【Method/process】Taking the three institutions
with the largest collaboration scale in the statistics discipline as an example, this paper constructs directed networks by collaboration
between supervisors based on the dissertation data under the joint postgraduate training and uses social network analysis to explore the evolution law and structural characteristics of networks from the macro and micro. Based on the exponential random graph model, the formation mechanism of collaboration is clarified from two aspects of network structure and supervisor's personal attributes.【Result/conclusion】The supervisors collaboration networks in the three universities have similar evolution trend, and show the characteristics of cluster, chain structure and multiple small community structure respectively in overall structure. The structure effect on the network formation is different among universities. Reciprocity and mediation promote the formation of collaboration, while closeness shows no effect on the formation of collaboration. There is no matching tendency in the research direction of supervisors, and attribute effects such as title of supervisor have a significant positive influence on network formation.【Innovation/limitation】This paper expands the research perspective of scientific collaboration and provides a reference for the discovery of collaboration laws and the exploration of the formation mechanism of collaboration between supervisors in joint postgraduate training. The applicability of the conclusion to other disciplines and special situations such as the inter-institutional and inter-disciplinary collaboration of supervisors remains to be investigated.