The spatial structure characteristics and influencing factors of knowledge networks are heated research topics in the field of economic geography. Previous studies in knowledge networks have focused on industrial cases, especially on knowledge networks at the national and city levels, while the research of individuals has been relatively limited. We firstly selected highly cited Chinese scientists based on 2014-2015 global highly cited scientists as the original data, and then established knowledge network based on the co-authorship of papers among those scientists, and finally applied social network analysis methods to examine the spatial structure of the knowledge network and used the negative binomial regression model to explore the influence mechanism of knowledge network from the perspectives of geographic proximity, social proximity and institutional proximity. The study found that: (1) the knowledge network of highly cited Chinese scientists has core-peripheral structure with the characteristics of a small-world network. (2) The spatial distribution of this knowledge network is dispersal at global scale and agglomerated at local level. (3) The formation of the knowledge network is influenced by the personal characters of those scientists. Geographical distance has a negative effect on the connection between scientists. The influence of geographic proximity still exists. Both social proximity and institutional proximity have significant positive effects on the formation of knowledge networks of highly cited Chinese scientists.

知识网络的空间结构特征与影响因素是经济地理学探讨的热点议题之一,以往研究侧重于产业案例的分析,主要关注国家和城市层面的知识网络,而对科学家等个人层面的网络研究较为缺乏。以2014—2015年全球高被引科学家为原始数据,筛选出高被引华人科学家,并基于Web of Science数据库,检索高被引华人科学家之间合著论文的数据构建知识网络,借助社会网络分析方法对高被引华人科学家知识网络的空间结构进行分析;并运用负二项回归模型,从地理邻近性、社会邻近性、制度邻近性3个维度,探讨高被引华人科学家知识网络的影响机制。研究发现：① 高被引华人科学家知识网络存在核心-边缘结构特征,且具有小世界网络的网络特征;② 此知识网络呈现“小集聚大分散”的空间结构特征,地理邻近性作用明显;③ 高被引华人科学家知识网络形成过程中会受到科学家自身科研能力等因素的影响,地理距离和科学家之间的联系呈现负相关关系,地理邻近性的影响仍然存在,社会邻近性和制度邻近性均对知识网络的形成有促进作用。

<p>Industry-university-research institute collaborative innovation process and its spatial structures attract the interest of researchers in many fields. With the rise of emerging economies and their technological upgrading, their spatial structure of innovation network is developing into an important research topic. And China, in particular, provides the opportunity to study the evolution of such network structures. With the help of some sophisticated data analysis software like SPSS, UCINET and ArcGIS, this paper discusses the above-mentioned issues based on graphical analysis and an empirical analysis of co-inventor networks of China's equipment manufacturing industry using patent data issued by the State Intellectual Property Office of P.R.China from 1985 to 2012. We reached three conclusions about the structural evolution of the industry-university-research institute collaborative innovation network of Chinese equipment manufacture industry. Firstly, our systematic examination has identified a rapid growth of patents and significant changes of actor composition from 1985 to 2012, which shows the rise of privately owned enterprises and universities around 2000, with universities standing out as the most significant and strongest actors in the process of building innovation networks, while state-owned enterprises only dominate some specific fields. Secondly, city-level is the major geographical scale of industry-university-research institute collaboration in developed cities; while undeveloped cities tend to cooperate with competent ones at provincial or national level. It is mainly because concentration of universities and firms with strong innovative ability makes it easy to find the perfect local partner, while weaker actors have to look for the best innovation partners across city boundaries. Last but not least, political decisions concerning R&D investment supported by provincial governments have a positive influence on interprovincial innovation activities. Meanwhile, the spatial political bias in China can lead to the hierarchical structure of Chinese innovation networks, which shows the significance of municipalities and provincial capital like Peking, Shanghai and Guangzhou.</p>

产学研合作是区域创新的主要途径和重要来源.以中国装备制造产业为例,基于中国知识产权局1985-2012年间的合作发明专利数据,借助SPSS,UCINET,ArcGIS等定量分析工具,对中国装备制造产业合作网络的创新主体结构,空间结构及其演变,创新合作的空间尺度的影响因素进行了分析.研究发现,民营企业,高校在中国装备制造产业创新网络中的地位不断上升,数量不断增加,且已经成为重要的创新源泉;市域空间合作成为发达地区城市产学研创新合作最重要的空间单元,国家空间是欠发达地区城市产学研创新合作的主要空间载体;理工科高校等科技资源的空间集聚态势是导致创新网络层级特征的主要因子,科技资源富集的行政中心如直辖市,省会城市等发达城市成为最重要的资源集聚地,创新源泉和创新合作对象.

The concept of megalopolis, since its original inception six decades ago, has inspired many new terms that mainly describe large-scale urbanized forms such as megaregions and polycentric urban regions. However, recent studies have increasingly focused on the two key functions that megalopolises act as an incubator of new ideas and trends and as a hub that articulates knowledge exchange at the megalopolitan, national, and global scales. While the recent studies have mainly analyzed the functional aspects of megalopolis based on China's Yangtze River Delta region, this paper investigates the evolving process and mechanisms of knowledge collaboration within and beyond Guangdong-Hong Kong-Macao Greater Bay Area (GBA) - one of the most promising and vibrant megalopolises in China. In addition, the GBA megalopolis is unique because it contains Hong Kong and Macao, which have a different political system from China's mainland. Drawing upon a dataset of publications that were indexed in Web of Science Core Collection during the 1990-2016 period, this paper uses the Gini coefficient to measure the degree of knowledge polycentricity of the GBA megalopolis. Here, knowledge polycentricity is further classified into attribute polycentricity of knowledge production and functional polycentricity of knowledge collaboration within and beyond the GBA megalopolis. Whereas the attribute polycentricity refers to the distribution inequality of the total publications of GBA cities, the functional polycentricity represents the distribution inequality of GBA cities' knowledge collaboration at different geographical scales. Our empirical results show: (1) knowledge production of the GBA megalopolis as a whole has experienced a robust and continuous growth. The degrees of both attribute polycentricity and functional polycentricity have also been on the increase in general, although there are some fluctuations in early years and some deviations in recent years. During the ten years after Hong Kong and Macao returned to China (the 2000-2010 period), the degree of knowledge polycentricity of the GBA megalopolis especially enjoyed the fastest rise; (2) The degree of functional polycentricity decreased with the expansion in the geographical scales at which it is measured, confirming the findings of previous studies that functional polycentricity is scale-dependent. Moreover, we find that the degree of functional polycentricity becomes more fluctuated at the global scale while it tends to increase continuously at the megalopolitan scale; (3) The evolving process of knowledge polycentricity of the GBA megalopolis is influenced by institutional proximity, geographical proximity and status proximity between cities. Specifically, the mobility of researchers, the collaboration of universities and research institutes, and the coordination of local governments are three major forces promoting the evolution of knowledge polycentricity of the GBA megalopolis. Overall, the increasing knowledge polycentricity would be of significance for the GBA megalopolis to form a knowledge-driven region of collective collaboration.

基于1990-2016年“Web of Science”核心合集所收录的科研论文合著数据,借助基尼系数测度属性和功能多中心性的方法,对粤港澳大湾区城市群的知识多中心性及其知识网络的演化进行了研究。结果发现：① 伴随着粤港澳大湾区城市群知识生产总量的持续增长,其多中心性程度呈现出阶段性、阶梯式提升的特征,分别经历了波动、增长和分化的发展阶段,港澳回归后的2000-2010年间是多中心性快速增长的重要阶段。② 粤港澳大湾区城市群在区域、国家和全球尺度上的功能多中心性程度随着尺度增加逐级递减,进一步证实了功能多中心性的尺度规律性;而且发现了多中心在演化中的尺度敏感性,全球尺度上的多中心性会存在不确定的峰值,而区域尺度上的多中心性可能会持续增加。③ 城市群多中心的演化是受制度接近、地理接近和等级接近影响,在研究人员移动、科研单位联动和政府政策推动及其行动主体间的相互作用下实现的,多中心程度的增加有助于推动粤港澳大湾区城市群构建科研协同创新共同体。

Based on the coauthored data of CNKI Core Journals and Web of Science Core Journals, and by means of UCINET and ArcGIS software, this paper studies how the 1000-Youth Elite in Yangtze River Delta colleges and universities influences the knowledge cooperation network in Yangtze River Delta region, and its characteristics analysis. The results are: there are great differences in Network centrality of knowledge cooperation in Yangtze River Delta region, and the region tends to have multiple cores; regions except the Yangtze River Delta region like HK, Taiwan, and Beijing are main gathering areas; the knowledge cooperation network co-established with overseas countries are North America countries like USA and European countries like the United Kingdom; in addition, the spatial pattern of knowledge cooperation network tend to remain the core-periphery structure; in addition, the influence of the 1000-Youth Elite in colleges and universities on cooperative network in Yangtze River Delta region will be studied from three aspects: proximity, urban energy level driving and social network driving.

基于2011-2017年CNKI与Web of Science核心合集所收录的论文合著数据，借助UCINET与ArcGIS软件研究高校“青年千人”对长三角知识合作网络影响及其特征分析。结果发现：长三角内部知识合作网络度中心性总体差异显著，呈现多核心趋势；长三角外部以香港、台湾和北京为主要集聚区；而与国外建立的知识合作网络则集中于以美国为代表的北美洲及欧洲的英国等国家；此外，知识合作网络在空间上呈现明显的“核心-边缘”格局并将长期存在。在此基础上，从邻近性、城市能级驱动、社会关系网络驱动等方面探讨高校“青年千人”对长三角区域知识合作网络的影响。

<p id="C3">Scientific collaboration has become an important part of the people-to-people exchanges in the Belt and Road Initiative, and remarkable progress has been made since 2013. Taking the 65 countries in the Belt and Road regions, hereafter called the BRI countries, as the research areas and using the collaborated WOS core collection papers to construct the inter-national scientific collaboration matrix, the paper explores the multiple structure and formation mechanisms of the scientific collaboration network of BRI countries through the GIS spatial analysis, social network analysis and negative binomial regression. The results show that: (1) at the global scale, knowledge flow is becoming more and more frequent, and the degree of internal collaboration within BRI countries is not as close as that of collaboration with other countries in the world. All the sub-regions of BRI regions have the most external collaboration links with Europe outside BRI regions. The global scientific collaboration network based on the BRI countries has changed from a "tripod" framework of Europe, USA, China to the "polygon" framework of Europe, USA, China, Japan, Canada, Australia and other dense regions. (2) The spatial structure of the collaboration network within the BRI countries is transformed from the "dual-core" type (China and Central & Eastern Europe) to the "multipole" type (Central & Eastern Europe, Russia, China, Singapore, India, Saudi Arabia, etc.). The hierarchical structure presents a typical "core-periphery" pattern, the core countries of which change from single core (China) to dual cores (China and India). In terms of control structure, the type has changed from one superpower (China) to multi great powers including China, India, Russia, Poland, Czech, Turkey, Malaysia and Iran, which presents a pattern of "high in the surroundings while low in the middle". (3) The main formation mechanisms of scientific collaboration network of BRI countries depend on scientific research ability, the level of economic development, the level of foreign linkage, and proximities such as geographical proximity, social proximity and language proximity. Among them, scientific research ability and social proximity play the most important role, while geographical distance gradually weakens the hindrance to scientific collaboration.</p>

科技合作是“一带一路”倡议“民心相通”的重要组成部分,2013年以来取得显著进展。以“一带一路”沿线65国为研究区,运用WOS核心集合论文合作数据构建国家间科研合作矩阵,通过GIS空间分析、社会网络分析以及负二项回归分析法,探究沿线各国2013年和2018年科研合作网络的多元化结构及形成机制,结果表明：① 在全球尺度,“一带一路”沿线各国与其他国家之间的知识流动日趋频繁,但内部合作紧密程度不及与沿线以外地区的合作程度。沿线区域对全球的科研合作网络由欧洲、美国和中国“三足鼎立”骨架结构向欧洲、美国、中国、加拿大、日本、澳大利亚等多个密集区构成的“多边形”骨架结构转变。② “一带一路”沿线地区的合作网络的空间结构,由中国、中东欧区组成的“双核发散”向“一区多点”构成的“多极化”空间结构转变;等级结构则呈现典型的“中心-外围”等级层次结构,且由单核心向双核心转变;控制结构呈现出由“单核心”向四周高、中间低的“多超多强”盆地型控制结构转变。③ 沿线国家的科研实力、经济发展水平、对外开放度等国家主体属性以及地理邻近性、社会邻近性等是沿线科研合作网络形成的主要机制,其中科研实力、社会邻近性的作用最大,地理距离对科研合作的阻碍作用逐渐减弱。

Taking ninety county-level cities in Zhejiang Province as geographical units, this article uses spatial analysis and negative binomial regression model to examine the structure, spatio-temporal evolution and impact mechanism of Zhejiang city innovative network from 2007 to 2017, with the data of co-application patents based on big text data mining. The study finds that: Firstly, the scale continues to expand and the structure becomes more and more complex in network evolution. In the study period, the scale of Zhejiang city innovative network has been growing from a sparse network to a dense network, but it is still underdeveloped. The network acts as small-world feature which is superior to random network, and the whole network has good accessibility and communication efficiency. Innovative network shows core-periphery pattern which is centralized, localized, and hierarchical, centers around Hangzhou Great Bay Area, leaves western Zhejiang cities in the edge. Secondly, as for impact mechanism, proximity of different dimension has different impact on the level of city innovative cooperation. In addition, city characteristics and network effects also have significant impacts. The intensity of city innovative cooperation in Zhejiang Province is significantly positively correlated with their economic scale, education level and policy support, and the technological gap and the administrative level also play a positive role. Geographical proximity and border proximity are positive factors affecting innovation cooperation. The relationship between cognitive proximity and innovative cooperation is u-shaped, which indicates that when cognitive proximity is at a low level, it is not conducive to innovation cooperation. However, with the increase of inter-city cognitive proximity, its marginal effect on innovative cooperation becomes more and more strong. On the contrary, relationship of innovative cooperation and institutional proximity shows an inverted ‘U-shaped’ curve, indicating that institutional proximity has a positive effect on innovative cooperation level at a lower level, but its effect will become weaker with the improvement of city market system. Technology proximity shows a positive effect, indicating that the similarity of technology structure makes the knowledge interaction among the cities more smoothly. Third, network effects play a positive role in promoting inter-city innovative cooperation. Compared with individual node network characteristics, network structure characteristics formed by cities surrounding the nodes play a more significant role in promoting cooperation. Innovative network more inclined to show the overall effect, so we should focus on the optimization of network structure, taking the advantage of role of core node.

以浙江90个县级城市（城区）为对象，基于大数据专利文本挖掘析取城市间合作申请专利数，构建创新网络，采用空间网络模型和负二项回归模型分析2007—2017年浙江城市创新网络的网络结构、时空演化及创新合作强度的影响机制。研究发现：① 浙江城市创新网络规模逐渐扩张，“小世界性”显著，通达性较好，整体呈以杭州湾区为核心的“网络局部化、辐射中心化”特征，等级层次性清晰；② 浙江城市创新合作强度与经济规模、教育水平、政策支持、技术势差、城市行政等级、技术邻近性、边界相邻效应显著正相关且受地理距离约束，认知邻近性和制度邻近性与创新合作间分别呈“U”型和倒“U”型曲线关系，网络效应更多在整体网层面上促进了创新合作。在特定省份县级城市层面探讨了如何加强城市创新网络协同效应，以促进地方城市间的创新联系。

As China enters the new normal era, the existing mode of joining into the global production network is confronted with the dual dilemma of internal dividend disappearance and external trade frictions. It is very important to explore the characteristics and driving forces of dynamic evolution of industrial linkages between China and other countries or regions in the world. Based on the world input-output table database from 1995 to 2014, this study depicts the global industrial interdependence network from the national scale using input—output analysis and social network analysis, and focuses on the evolution of China's role and position in this pattern using spatial econometric model. The main findings are as follows: China has become a bridge between East Asia and Southeast Asia and other emerging markets, thus upgrading from an "outsider" to an important hub in the industrial network linking Europe and the United States. However, from 2015 to 2019, the expansion of China's foreign industrial linkages has gradually reached a plateau. From the perspective of supply side, China's resource and manufacturing industries are constantly embedded in the supply side of the global production network. However, the overall external supply capacity of most China's service industry is relatively below global average. In terms of demand, China, as the "factory of the world" and a major country in infrastructure construction, has an increasing global influence in manufacturing and infrastructure. Nevertheless, the international influence of productive services like Transportation, Logistics and Finance has been declining since 2005. The Chinese industries obtain the local added value overall to continue to grow, but the efficiency of the added value capture is low. On this basis, this study explores the influence of multi-dimensional proximities and trade protectionism on China-global industrial linkages and their value added. Among them, the geographical and cognitive proximity has the most extensive influence, while the social and institutional proximities only promote and strengthen the intermediate production linkage with more complex forms of industrial linkages and final demand linkages. However, technical barriers and anti-dumping investigation have weakened China's foreign industrial links. The sanitary and phytosanitary measures have a significant weakening effect on the final demand linkage, but not on the intermediate product production. To sum up, the upgrading of production-oriented service industry and the efficiency of the acquisition of overall added value are the potential directions for the expansion of China's foreign industrial linkages, while the construction of domestic industrial cycle and industrial diversification are the reasonable measures for the expansion of China's foreign industrial linkages.

随着中国经济步入新常态,中国嵌入世界生产网络的既有模式遭遇内部红利消失与外部贸易摩擦的双重困境,探讨中国同世界其他国家/地区间的产业联系特征及其动态演化机制,对寻找中国产业对外联系的破局方向至关重要。基于1995&#x02014;2014年世界投入产出表数据库,从国家尺度刻画世界产业相互依赖网络,并着重关注中国在这一格局中的角色与地位演变。研究发现：① 1995&#x02014;2014年中国从世界生产联系网络的边缘国家演变成为沟通东亚、东南亚地区与其他新兴市场的桥梁,并从美欧主干联系的&#x0201c;局外人&#x0201d;升级成为链接欧美产业网络的重要枢纽。而2015&#x02014;2019年中国对外产业联系拓展逐渐进入曲折发展的瓶颈期。② 从供给角度看,中国基础资源行业和制造业正不断嵌入世界生产网络的供给侧。相比之下,中国大部分服务业对世界产业网络的供给能力低于世界平均水平。③ 从需求看,中国作为&#x0201c;世界工厂&#x0201d;与基础设施建设大国,在制造业与基建方面具有世界性的影响力,然而中国的交通、物流、金融等生产型服务业的国际影响力在2005年后逐步下降。④ 中国产业获取本地附加值总体持续增长,但附加值捕获的效率较低。在此基础上,本文探讨了多维邻近性、贸易保护对中国&#x02014;世界产业联系的作用,发现：中国对外产业联系的演化受地理、认知、社会与制度4个维度的邻近性影响并形成路径依赖。其中地理与认知邻近性的影响最为广泛,而社会与制度邻近性仅对产业联系形式更加复杂的中间生产联系与发展较为成熟的最终需求联系有促进作用,技术贸易壁垒与反倾销调查会削弱中国对外产业联系;卫生安全检疫措施对最终市场需求联系有显著削弱作用,而对中间产品生产联系的作用并不显著。综上所述,生产型服务业与总体附加值获取效率是中国对外产业联系拓展的潜力方向,而内销&#x02014;出口平衡与多元化的产业发展策略是中国对外产业联系拓展的合理举措。

The research of innovation networks is a hot topic in innovation geography, and the evolution mechanisms of innovation networks are one of the core issues in the research of innovation networks. Based on CiteSpace knowledge map and deep literature analysis, this article summarizes the international research on the evolution mechanisms of innovation networks. The results show that: The endogenous effects of networks, including the overall characteristics, embeddedness, externality, technology gatekeeper, and absorptive capacity of networks, mainly reflect the evolution of innovation networks in terms of the degree of dependence of organizations on local areas, the absorptive capacity of external knowledge, and the ability to control overall networks. The characteristics of network organizations, which are embodied in the scale, nature, and size of different organizations, also have effects on the evolution of innovation networks. Multidimensional proximities play an important role in the evolution of innovation networks, and the single proximity and multidimensional proximities have periodic characteristics for the evolution of innovation networks. The evolution mechanisms of innovation networks differ at various stages of development, and the combination modes among mechanisms are also different. Finally, based on the analysis of the contents and methods of international research on the evolution mechanisms of innovation networks, the shortcomings of current research and directions for future research are put forward.

创新网络研究是创新地理学研究的热点,其中创新网络的演化机制是当前创新网络研究的核心问题之一。论文通过CiteSpace知识图谱定量分析和文献深度分析相结合,对国外创新网络演化机制的研究进行全面解析,结果表明：① 网络内生效应,包括网络的整体特征、根植性、外部性、技术守门员与吸收能力等,对创新网络的演化主要体现在组织要素对当地的依赖程度、对外部知识的吸收能力,以及对整体网络的控制能力;② 网络组织要素的规模、性质和大小等对创新网络的形成演化也有一定的作用;③ 多维邻近性对于创新网络的形成和演化具有重要作用,单一邻近性和邻近性之间的综合作用对于创新网络的演化具有阶段性特征;④ 不同发展阶段创新网络的演化机制不同,机制之间的组合模式也存在差异。最后,在总结国外创新网络演化机制研究的主要内容和方法的基础上,提出了当前该领域研究中存在的不足及未来研究的方向。

With the growing importance of University-Industry cooperative innovation (U-I cooperative innovation) in regional innovation, there is an increasing concern over the factors influencing U-I cooperative innovation. While U-I cooperative innovation features a process of knowledge transfer, multi-dimensional proximity is an appropriate analytical perspective to study the influencing factors of U-I cooperative innovation. This paper argues that geographical proximity, cognitive proximity and social proximity are essential elements of a conceptual framework for the analysis of U-I cooperative innovation, which contains heterogeneous organization. As such multi-dimensional proximities represent an important factor in the promotion of U-I cooperative innovation, interactive learning has been proved to be the way to realize knowledge transfer under the influence of multi-dimensional proximities for cooperative subjects—since frequent and continuous interaction between U-I cooperative subjects can enhance the level of U-I cooperation. By constructing a theoretical framework of "Multi-dimensional proximities, geographical proximity and related proximities→Interactive learning→Level of cooperation", and based on a multi-case study methodology, this research takes high-tech enterprises in the Guangzhou Development District as an example and explores the influence of multi-dimensional proximities on "point to point" U-I cooperative innovation. The research findings show that: (1) while geographical proximity, cognitive proximity and social proximity all contribute to the level of U-I cooperation, these positive effects vary at different stages of technological innovation; (2) whilst interactive learning has significant moderating effects on multi-dimensional proximities and the level of U-I cooperation, there are noticeable periodic characteristics in its effects in terms of content, way and intensity; (3) geographical proximity, cognitive proximity and social proximity, respectively, have complementary and substitutive effects on the level of U-I cooperative innovation, although the effects may vary in different stages. In the interaction between different types of proximities, the positive influence of complementary effects is usually greater than that of substitutive effects. The conclusion is useful for us to understand the interaction process between the innovation subjects under different circumstances of proximities; in addition, it can also provide evidence for policy-making as regards rational distribution of scientific and technological resources, selection of U-I cooperative partners, as well as appropriate responses to different circumstances of proximities in the process of technological innovation cooperation.

多维邻近性是研究产学研合作创新影响因素的恰当的分析视角。构建“多维邻近→互动学习→合作程度”理论分析框架,运用多案例方法研究多维邻近性对项目形式的产学研合作创新的影响。研究表明：① 地理邻近、认知邻近、社会邻近对产学研合作程度的提升均有积极影响,但在技术创新的不同阶段存在差异。② 互动学习对多维邻近与产学研合作程度具有显著的调节作用,在内容、方式、强度上有明显的阶段性特征。③ 地理邻近、认知邻近、社会邻近对产学研合作程度的交互影响呈互补效应或替代效应,在特定情况下存在阶段性差异;互补效应的积极影响通常优于替代效应。

With the rise of the knowledge-based economy in the 1980s, knowledge (including code and tacit knowledge) as the backbone of innovation has become a key factor affecting production process. Cities have gathered not only a large number of professionals, universities and research institutions, but also a great many producers and consumers, which provides the premise for the innovation actions. City's knowledge storage and its position in the regional knowledge network play an important role in comprehensive competitiveness. Published papers and patents are main outcomes of innovation, which are used to evaluate the urban innovation capability. Moreover, co-publications and co-patents are not only the form of knowledge spillover, but also the key indicators to measure regional innovation. Taking the co-publication and co-patent in the field of biotechnology in China during 2000-2009 as the original data, we built scientific knowledge network (SKN) and technological knowledge network (TKN) between cities. From the perspective of complex networks and geospatial analysis, we explored the temporal and spatial complexity of knowledge spillovers combining the indicators of whole network structure, ego network, power-law, hubs and so on. The results show that: firstly, the nodes degree distribution of SKN and TKN is consistent with the power-law distribution, which means that the both networks not only have a scale-free network structure, but also present a preferential attachment rule when the cities choose the cooperation partner. Secondly, central cities have an obvious hierarchical structure, and are featured by a "big scattered and small gathering" spatial pattern in SKN, while the TKN is not showing this feature. From the view of central city ego network, the cooperation develops between the coastal capital cities at first, and then turns to inter-regional cooperation, such as Yangtze River Delta, Pearl River Delta, and inter-regional knowledge spillovers is obvious in SKN. The central cities and its partners are still in the coastal city instead of western provincial capitals, and inter-regional knowledge spillovers are not significant in TKN. Thirdly, the temporal evolution of central cities and its ego-network presents hierarchical diffusion and contagious diffusion, and conforms to law of grades process in SKN. The TKN is dominated by hierarchical diffusion. Finally, this study draws conclusions on the temporal and spatial complexity of innovation network, which has a positive impact on quantifying spatial knowledge spillovers and measuring its space-time evolution. Besides, the results clarify the status of each city in innovation networks, which provides a new perspective for the cities to formulate innovative policies.

知识在产业集聚、区域创新中的地位越来越突出,城市知识储量及其在区域知识网络中的地位对城市的综合竞争力有重要影响。学术论文合作与专利合作是知识溢出的体现形式,是科学和技术发展的重要成果,也是度量区域创新能力的主要指标。以2000-2009年中国生物技术领域合著论文和共同申请专利的信息为原始数据,分别构建中国城市间科学知识网络(scientific knowledge network,SKN)与技术知识网络(technological knowledge network,TKN)。运用复杂网络与地学空间分析方法,从整体网络结构特征、择优链接性、中心城市及其自我网络的空间特征等方面进行分析,探究知识溢出的时空复杂性。研究表明：①SKN和TKN具有无标度网络特征;SKN节点度数的异质性高于TKN。②两种网络均呈异配性,即城市选择合作对象存在明显择优链接性,知识溢出具有粘着性和空间依赖性。③SKN中心城市具有明显的等级结构,空间分布总体呈&#x0201c;大分散小集聚&#x0201d;特点;TKN中心城市层级结构不明显,尚未形成明显极化中心。④SKN中心城市自我网络的合作空间,由最初的沿海省会城市间的合作转向长三角、珠三角、京津冀等区域间和沿海城市与内陆城市间的合作,区域间知识溢出明显;TKN中心城市自我网络仍多分布于沿海城市和少数中西部省会城市,区域间知识溢出不明显。⑤SKN中心城市及其自我网络的时空演变存在等级扩散和传染扩散的现象,符合时空梯度推移规律,且空间等级梯度逐渐向扁平化转变;TKN中心城市及其自我网络的时空演变以等级扩散为主,时空梯度推移现象不明显。研究结论为量化知识溢出及知识溢出网络结构的时空演化过程提供新的研究视角,对城市创新政策的制定有一定借鉴意义。

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.

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。

Over the past decade, there has been an explosion of interest in network research across the physical and social sciences. For social scientists, the theory of networks has been a gold mine, yielding explanations for social phenomena in a wide variety of disciplines from psychology to economics. Here, we review the kinds of things that social scientists have tried to explain using social network analysis and provide a nutshell description of the basic assumptions, goals, and explanatory mechanisms prevalent in the field. We hope to contribute to a dialogue among researchers from across the physical and social sciences who share a common interest in understanding the antecedents and consequences of network phenomena.