基于随机行动者模型的全球科学合作网络演化研究

在知识经济时代,国家（地区）之间的科学合作日益频繁,科学活动的全球化和网络化特征越来越突出。国家（地区）间的合作不仅仅取决于双边的关系特征,还受到网络内生的结构效应影响。基于2000—2019年国家（地区）间的科学论文数据,本文采用社会网络分析方法和随机行动者模型,探讨全球科学合作网络的演化态势与影响机理。研究发现：① 全球科学合作网络规模日益扩大,凝聚性不断增强,具有明显的小世界性和部分的无标度特征,呈现去中心化的趋势。② 全球科学合作网络的拓扑结构呈现等级层次式与分布式并存的组织特征,中国逐渐从边缘向中心靠近,网络结构由美国单核演变为中美双核。③ 回归结果表明以传递性和择优连接为主的结构内生性是网络演化的首要驱动因素,以地理邻近性和认知邻近性为代表的国家（地区）间邻近性是网络演化的关键影响因子,国家（地区）规模也是促进合作网络演化的重要因素。与此同时,本文还发现传递性、地理邻近性、认知邻近性和国家（地区）规模的重要性日益提高,择优连接性对网络演化的影响逐渐减弱。

The evolution of the global scientific collaboration network: A stochastic actor-oriented model approach

In the age of globalizing knowledge economy, inter-country scientific collaborations are more and more frequent. Therefore, the globalization and networking characteristics of scientific research activities are increasingly prominent. The scientific collaboration between two countries does not only rely on the bilateral relationship between, but also is influenced by the structural effects of the network itself. However, the endogenous structural effects are ignored in the literature on the evolution of knowledge network. Based on the internationally collaborative papers from the Clarivate Analytics' InCites database in the period 2000-2019, this study uses social network analysis and stochastic actor-oriented model to explore the structure, dynamics and determinants of global scientific collaboration network. Results show that the size of the network has expanded, the number of nodes and ties in the network has substantially increased, and the densification of the network has strengthened over time. Distinctive characteristics of the network is typically small-world nature and partially scale-free distribution, and there is a significant trend towards decentralization in the collaboration network. The whole network displays the co-existence of hierarchical “star-shaped” structure and heterarchical “universal” structure. China moves up from the periphery to the core, and the network is evolving from a single-center dominated by the United States to the double center including Sino-US, the bilateral partnership between China and the United States becomes the most importantly bilateral collaboration in the world. In addition, stochastic actor-oriented model indicates that transitivity and preferential attachment positively drive the evolution of scientific collaboration network. Geographical proximity and cognitive proximity have a positive and significant effect on the formation of international collaboration. The dynamic mechanism of the global scientific collaboration network is facilitated by country size. Meanwhile, common language, post-colonial links, and international students play an important role in the dynamics of global scientific collaboration network. Besides, we find that the effects of transitivity, geographical proximity, cognitive proximity and country size have increased, while the impact of preferential attachment has waned over time.