中国城市知识合作网络结构演化的影响机制

The influencing mechanisms of evolving structures of China’s intercity knowledge collaboration networks

城市知识合作网络的影响机制是创新地理的重要研究议题,已有研究大多关注外生动力(城市资源禀赋、多维邻近性等)的影响,而较少关注内生动力因素(择优链接、传递性等)。基于“Web of Science”中的论文合作发表数据构建中国城市知识合作网络,分析其2006—2016年空间结构和拓扑结构的演化特征,并通过加权随机指数图模型(ERGM)定量揭示内生和外生动力对该网络形成的影响。结果显示:①从空间结构看,东密西疏的格局比较稳定,但整体由北京—上海双核结构向多中心发展,五大国家级城市群成为孕育多极的重要空间。以高能级城市为核心形成的节点区域愈发与规划的城市群范围一致,其发挥着知识资源邻近溢出和远程交互的双重作用。②从拓扑结构看,知识合作的规模和密度显著增加,对大城市的择优选择效应明显。但随着多中心崛起,网络极化程度和异配性均在弱化;随着城际合作路径不断优化,网络聚合性提升,成为小世界网络。③从影响机制看,自演化与择优链接是知识网络的重要驱动力,其作用与城市等级效应相重叠;城市知识规模属性对知识流动的正向影响弱于多维邻近性和路径依赖性;高铁的存在有利于促进知识合作,而地理距离的影响不显著。

The study of influencing mechanisms of intercity knowledge collaboration networks is an important research topic of innovative geography. Most existing research focuses on the influence of exogenous forces(e.g., urban resources endowment, multidimensional proximity),whereas less attention has been paid to endogenous factors(e.g., preferential attachment,transitivity). This research constructs an intercity knowledge collaboration network of Chinese cities based on the co-publication data from the Web of Science, analyzes its evolving characteristics of spatial and topological structures from 2006 to 2016, and quantitatively explores the endogenous and exogenous forces underlying the network formation through exponential random graph modelling. The results show that:(1) From the spatial structure perspective, the intercity knowledge flows are dense in the eastern region but spare in the western region, which is stable during 2006-2016. The overall network has developed from a dual-core structure of Beijing and Shanghai to a polycentric structure, in which five nationallevel urban agglomerations have become important bases for nurturing multiple centers. The nodal regions centered on highly administrative cities have become increasingly consistent with the planned urban agglomeration, which plays the dual role of proximal spillover and remote interaction of knowledge resources.(2) From the topological structure perspective, the scale and density of intercity knowledge flows have increased significantly, and the preferential attachment to big cities is obvious. However, with the rise of multiple centers, the network polarization and disassortativity have been weakened. With the optimization of intercity knowledge collaboration paths, the network cohesion has improved, thus becoming a smallworld network.(3) From the influencing mechanism perspective, self-evolution and preferential attachment are important driving forces of knowledge collaboration networks,showing an overlapping effect with urban hierarchy. The positive impact of urban knowledgerelated attributes on intercity flows is weaker than multidimensional proximity and path dependence. The presence of high-speed railways promotes knowledge collaboration, while the influence of geographic distance is not significant.