The Network Structures and Characteristics of Global Hubs of STI in Big Data Analysis and Mining Technology

Abstract

Global hubs of STI are strategic nodes in the global science network and will lead the future of global scientific progress. This paper takes big data analysis and mining technology as an example, exploring the structures and characteristics of global science network of 15 representative global hubs of STI from 2000 to 2019. Based on the data of highly cited papers and methods of social network analysis, we find that the global science network has the structure of hierarchy, with an obvious characteristic of regionality. Based on the structures and characteristics of global science network, the knowledge flow and management model of global hubs has been proposed, which may be conducive to strengthening the knowledge management and network collaboration among global hubs of STI in other frontier technologies.

Keywords

Global science network global hubs of STI social network analysis big data analysis and mining technology

Introduction

Global hub of STI is a city or a special region that occupies a leading and active position in global science, technology and innovation, with powerful influence to guide, organise and control the flow of global resources. As an important source of new knowledge, the global hub of STI is a critical engine for global technological innovation and economic growth (Bai et al., 2020). Basic research ability is an important indicator to measure the level of comprehensive strength of a region (Busher, 1976), which is also the focus of the global hub of STI.

The interaction of global hubs of STI will contribute to generating new knowledge that may lead the world in change. As the cluster of the world’s leading universities, research institutes and enterprises, the global hub of STI is rich in knowledge with the outstanding ability of basic research. The interaction of global hubs of STI will affect the flow and agglomeration of basic science resources such as researchers and programmes, accelerate the recombination of knowledge elements, guide the global cooperation and competition in the specific science area, and ultimately change the progress of global frontier technologies (Debin & Shunhui, 2016). During this process, the global science network has gradually formed and expanded, reflecting the impact degree and interactive mode among the global hubs of STI. However, few research has objectively described the interaction of global hubs of STI from the perspective of global science network (Chaminade & Plechero, 2015; Ma et al., 2022).

The global science network refers to the cooperation and exchange of science among different countries or regions, the sharing of science resources, and the formation of a network that gives full play to the comparative advantages of basic research (de Beaver & Rosen, 1979). Constructing the global science cooperation network and the global science citation network is beneficial to objectively describe the interactive mode of global hubs of STI. It helps to explore the influential hubs in the network and the approaches to forging links between different hubs. More importantly, the global science network can reflect the flow and creation of knowledge, which is of great significance for the global hubs of STI to strengthen knowledge management and carry out original basic research.

In the digital economy, big data analysis and mining technology is the core and basic technology of the artificial intelligence industry, and it is a strategic technology leading the future. The world’s major developed countries regard this technology as an important strategy to enhance national competitiveness and maintain national security, then step up the introduction of plans and policies. Selecting a representative technology rather than all technology areas to analyse the global science network of global hubs of STI is conducive to understanding the interaction topics in specific technology areas, then detecting the knowledge factors affecting the structures and characteristics of global science network. The problem, however, is that global hubs of STI with comprehensive research ability may not be reflected in the global science network.

Accordingly, this study takes big data analysis and mining technology as an example, exploring the structures and characteristics of the global science network of 15 representative global hubs of STI from 2000 to 2019. Based on the data of highly cited papers from the WOS database and methods of social network analysis, the structure of hierarchy has been found in global science network, accelerating the flow of new knowledge. Furthermore, we find that global science network has an obvious characteristics of regionality, which does not coincide with its global property. Based on the structures and characteristics of the global science network, we propose a knowledge flow and management model of global hubs, to strengthen the knowledge management and network collaboration in global science network, which may have some recommendations for knowledge management of other frontier technologies in the world.

In summary, we mainly focus on the network structures and characteristics of global hubs of STI, instead of comparing the influence or status among them. To answer this question, the rest of this paper is organised as follows: Firstly, literature review is conducted to conclude research progress and gaps; then, we discuss the data and methods used in our research; after that, this paper explores the network structures and characteristics of global hubs of STI in big data analysis and mining technology; based on visual analysis, a knowledge flow and management model is proposed to improve the efficiency of knowledge creation for global hubs; finally, conclusions and discussions are provided.

Literature Review

Global Hubs of STI

Cities are important spatial carriers of regional economic and social activities. In the new era of global knowledge-intensive capitalism (Yun, 2022), cities are becoming focal points for knowledge learning and creation (Krishna, 2017). These learning cities function as collectors and repositories of knowledge and ideas and provide the underlying environment or infrastructure which facilitates the flow of knowledge, ideas and studying. The urban development mode gradually changes from increasing factor input to improving total factor productivity, followed by the increasing contribution of innovation to urban economic development (Ran & Liu, 2023). Current global cities are in the process of transforming from investment-driven cities to innovation-driven cities (Jian & Qiyu, 2016); therefore, many countries are exploring the construction of innovative cities with global influence. New York¹, London², Tokyo³, Singapore⁴, Seoul⁵ and other cities have proposed to build innovative cities with high innovation capacity in their long-term development strategies for 2030 and 2050.

In the Human Development Report (2001), the United Nations Development Program evaluated the technological innovation capabilities of various countries and put forward the concept of the global hub of STI for the first time. In this report, the global hubs of STI are clearly defined based on four dimensions: the ability of area universities and research facilities to train skilled workers or develop new technologies, the presence of established companies and multinational corporations to provide expertise and economic stability, the population’s entrepreneurial drive to start new ventures and the availability of venture capital to ensure that the ideas make it to market. In essence, these concepts are focused on the field of science and technology, emphasising the important role of scientific intersections in the development of science and technology in organisations, regions and countries. But there is no consensus about the definition of global hubs of STI.

Debin and Shunhui (2016) explained that the essence of global hubs of STI refers to the concentration of R&D investments, research intensive universities, innovation policies and technology achievements. Junjie et al. (2019) believed that the global hubs of STI are a few cities with high-level science and technology innovation in the world. Jianwen et al. (2015) pointed out that the global hubs of STI are the centres with intensive resources of science and technology, active innovation culture and atmosphere, which are conducive to drive urban agglomeration. Yun (2022) believed that the future global hubs of STI would be comprehensive innovation centres with a high degree of integration of science and technology, economy and culture, creativity and entrepreneurship.

Based on the previous studies, we propose that the global hub of STI (referred to as ‘global hub’) is a city or a special region that occupies a leading and active position in global science, technology and innovation, with powerful influence to guide, organise and control the flow of global resources. It is no doubt that global hubs include multiple innovation subjects, factors, processes and outputs, whereas the ability of science breakthroughs and knowledge creation is the hubs’ core and engine. Delu (2014) pointed out that the global hubs evolved from the science centre in the world. Raspe and Van (2006) regarded the knowledge stock and efficiency of transmitting knowledge to economic applications as crucial factors in explaining the cities’ competitiveness. Yun et al. (2015, 2019). thought that the breadth and depth of open innovation will increase, as a city changes from an industrial city to a knowledge city. Due to the importance to the worldwide innovation and productivity, global hubs have to strengthen the management and interaction of science and knowledge in global science network, which is also the main concern of our research.

Global Science Network

As an activity, science characterises the practical activities of human exploration of nature and laws (de Beaver & Rosen, 1979). Katz and Martin (1997) pointed out that cooperative research is the research work carried out by researchers in order to jointly produce scientific knowledge. As the globalisation of science and technology continues to deepen, the interconnections among the innovation systems of various countries continue to strengthen (Krishna, 2022), and the global science network develops rapidly. Innovation entities in the network are still consolidating and expanding innovation partnerships, accelerating the flow of innovation elements and interaction among innovation entities, and making full use of global resources (Bai et al., 2020).

Global science cooperation is an important form of scientific cooperation in the age of technological globalisation. According to the general concept of the cooperation network, global science cooperation is the scientific research personnel or organisations from two or more countries or regions to jointly create new scientific knowledge for common research goals. Judging from the address of the author’s institution in the paper, if a paper is co-signed by authors of two or more countries or regions, it is considered that international science cooperation has occurred and a global science network has been formed. The global science network refers to the cooperation and exchange of science among different countries or regions, the sharing of science resources, and the formation of a network that gives full play to the comparative advantages of basic research (de Beaver & Rosen, 1979).

With the development of the global science network, cities with better internationalisation have become the core node cities in the global science network by continuously accumulating R&D institutions and other innovative and dynamic organisations. Node cities are the most concentrated places for innovation capital in network organisations. Node cities have become the main support carriers for the construction of global science networks, and have shown strong competitiveness. How to search for knowledge on a global scale and use it for oneself, and establish connections with the world is the main research content of the global science network. The primary tasks of the core node cities are also reflected in the two aspects of searching for knowledge and establishing connections with the world.

There are still few direct studies on the relationship between global hubs of STI and global science networks. In the field of regional research, Clark et al. (2018) regarded cities as platforms for resource integration and believed that cities play a central role in regional scientific research and technology communication networks. Unlike Clark et al. (2018), Scott (2001) proposed the concept of ‘global urban area’ and Soja (2015) proposed the concept of ‘regional urbanisation’. They consider cities to be part of a region and emphasise the relationship between cities in local and regional networks. Wagner et al. (2015) studied the regional characteristics of the global science cooperation network from the perspective of global cooperation papers. He found that due to the rapid increase in research and development, cities in emerging countries such as China are increasingly likely to participate in and attract science cooperation in other regions, becoming an important node in the global science network. In the field of regional economy, Matthiessen et al. (2001) determined the interrelationships between different types of cities in the science cooperation network and the factors influencing them based on the cooperation and citation relationships among the top 40 cities in the world.

Big Data Analysis and Mining Technology

Big data represents a new technology paradigm for data that are generated at high velocity and high volume, and with high variety (Lee, 2017). Big data analysis and mining technology are important basis for exploring new knowledge and creating new value from big data (Li, 2016). In order to build smart cities, many countries have introduced big data analysis and mining technology into urban construction and development and demonstrated feasibility, generality and effectiveness of Urban Data Science (Balduini et al., 2019). Big data analysis and mining technology have wide applications for urban transportation (Bachechi et al., 2022), energy efficiency (Arif et al., 2022), health protection (Imran et al., 2021) and so on. Therefore, as global hubs, the basic research of big data analysis and mining technology has been placed in an unprecedented position (Balduini et al., 2019). Increasing international competitiveness of big data analysis and mining technology has become a strategic choice of global hubs.

In big data analysis and mining technology, Previous research discussed global science networks based on the level of nation, institution and author. At the national level, Li (2016) analysed the spatial pattern and cooperation network of major countries by taking the number of research papers as the measurement index. It is found that research centres are presenting a ‘core-edge’ trend, and the research strength and influence of the main developed countries in the core area are growing steadily, while the output of research results in India and China is growing fast. At the institutional level, Ying et al. (2015) found that the cooperation of big data analysis and mining technology was limited to universities, while the cooperation was not active among enterprises and other application-oriented institutions. At the author level, Xiaozan et al. (2021) explored the influential authors and internal relationships, as well as revealed the network structure among nations and institutions in scientific research related to big data analysis and mining technology.

However, there is very few research on the global science network of big data analysis and mining technology at the city or regional level, especially from the perspective of global hubs. Ma et al. (2022) investigated the formation mechanism of big data analysis and mining technology cooperation networks by considering the combined effect of three key factors, that is, the individual characteristics, relationship characteristics and cooperation characteristics of R&D entities instead of global hubs. Guan et al. (2015) constructed collaboration network at the city as well as country levels and found that inter-country collaboration network structure moderates the relationships between inter-city collaboration network structure and innovation performance, but didn’t discuss the global science network of global hubs in big data analysis and mining technology.

In general, current research about global hubs in big data analysis and mining technology is mainly focused on the application areas of this technology and only a few studies have been conducted from the perspective of global science networks. Furthermore, the research related to global science network of data analysis and mining technology is discussed at the level of nation, institution and author, but the role and function of global hubs have been largely ignored. However, global hub is a city or region that occupies a leading and active position in global science, technology and innovation, with powerful influence on the formation and evolution of big data analysis and mining technology. As a result, it is important and necessary to explore the global science network of global hubs in big data analysis and mining technology.

One major contribution of this paper is that we extract the information of global hubs based on the address of authors’ institutions, visualise the global science network of global hubs in big data analysis and mining technology and finally construct the knowledge flow and management model. It may be conducive to strengthen knowledge management and network collaboration among global hubs in other frontier technologies.

Data and Methods

Data

Database selection

Research papers reflect the latest scientific trends and contain valuable information about authors, institutions and addresses, while also providing a solid foundation for collaborative research (Bai et al., 2020). By analysing this information through network analysis methods, we can obtain important information about the growth of international R&D cooperation, the structures and characteristics of the global science networks.

We select Web of Science Core Collection (WOSCC) as the target database to provide the data about research papers on big data analysis and mining technology. WOSCC is an influential academic literature index database in the world, including 10 sub-databases such as SCI, SSCI and ESCI. The database contains more than 21,000 authoritative and influential academic journals around the world, covering natural science, engineering technology, biomedicine and other fields. WOSCC has a strict screening mechanism to ensure the quality of publications. According to Bradford’s law in bibliometrics, it only includes important academic journals and international conferences in various disciplines.

Search Strategy

Based on expert opinions from multiple rounds of search and discussion, this paper draws on the keyword search strategy to finalise the field. The logic operation rule is based on the logic operation rules of the Web of Knowledge and uses the logical connectors (OR, AND) to determine the paper retrieval strategy. The Web of Science Database was selected for data retrieval and data download. Search keywords of big data analysis and mining technology are as follows: TS=((big data or large data or mass data) AND (data analys* or data min* or DM or data analy* or statist* analysis or Machine Learn* or machine study or machine-learn* or semi-supervised learn* or integrated learn* or transfer learn* or probabilistic graphical model or PGM or bayesian network* or decision tree or support vector machine* or SVM or markov* model* or hidden markov model* or neural net* or KNN or k-Nearest Neighbor or time series analysis or sequence analysis or cluster* or rough set or autoregressive model or regressive model* or R language or real-time analysis or real-time analysis or real-time analysis)).

Data Download

Through preliminary data retrieval, the earliest paper in this field appeared in 1974. There are 54,116 papers in total. To ensure the comprehensiveness of the analysis, this article is downloaded according to the top 20% of highly cited papers each year. The period of data download is from January 1, 2000 to December 31, 2019. It can be found that the number of highly cited papers on big data analysis and mining technology before 2000 was less than 10. After 2019, global basic research has been inevitably affected by COVID-19, international politics, global trade and other factors, which is not conducive to describing the network structures and characteristics of global hubs objectively and naturally. Based on the improved word frequency query method, according to the search formula determined above, the original data with 9,833 research papers is downloaded.

Crawler Technology

To visualise the global science network of global hubs in big data analysis and mining technology, we extract the information of global hubs based on the address of authors’ institutions by crawler technology. For the authors’ institutions that cannot be effectively identified by crawler technology, we manually collect the cities’ addressess where the authors’ institutions are located, then clean and summarise the information of global hubs.

Data Screening and Collection

According to the institution of the author of the paper, the city or region can be traced back. The preliminary data screening can be used to calculate the top 30 places in the field of big data analysis and mining. However, not all regions can be called global hubs of STI, for the basic research status of one technology cannot represent the comprehensive strength of a region. Therefore, this article compared the top 30 regions based on the ‘Global hubs of STI Evaluation Report 2020’ published by Shanghai Information Center. To minimise the deviation of evaluation, this report takes nearly 150 major innovative cities or metropolitan areas around the world as the evaluation object, and focuses on displaying and analysing the top 100 cities in the comprehensive score for four consecutive years. It selects 23 indicators in four categories, namely basic research, industrial technology, innovative economy and innovation environment, with a high degree of reference significance.

After matching cities or regions that meet the above two standards at the same time, we finally establish a total of 15 global hubs in big data analysis and mining technology, namely Beijing, Silicon Valley, Boston, Hong Kong, Sydney, London, Washington, Shanghai, Melbourne, New York, Los Angeles, Singapore, Seoul, Zurich and Paris⁶. The data generation process of 15 global hubs has been summarized in Figure 1.

Figure 1.

Source: Summarized by authors.

Methods

The analysis method of this article is social network analysis, and the software used is ITG. ITG software is a scientific text mining and visualisation analysis system, which integrates natural language understanding and visualisation technology. It can be widely compatible with patent data of SCI, EI and other countries, and it is an intelligent technical term extraction software (Yuqin, 2015). Social network analysis describes the overall characteristics of the global science cooperation network, and centrality analysis is used to measure the importance of nodes in the network structure (Wolfe, 1995). Centricity indicators include individual actors’ centricity and the whole network’s centricity. In social network analysis, degree centrality, betweenness centrality and closeness centrality are relatively common indicators (Marcus et al., 2006).

Degree centrality is an indicator of the importance of a node in the network. This indicator can be divided into absolute centrality and relative centrality (Freeman, 1991). By calculating the total number of nodes connected to all other nodes, as shown in Equation (1):

C_{D} (N_{i}) = \sum_{j = 1}^{g} X_{i j} (i \neq j)

(1)

where C_D (N_i) represents the absolute degree centrality of node i, Σ^g_j=1 X_ij is used to calculate the number of direct connections between node i and j (i ≠ j, excluding i’s contact with itself).

In order to eliminate the impact of network size on the centrality of degrees, Freeman (1991) proposed a standardised formula, the relative centrality can be expressed as Equation (2):

{C^{'}}_{D} (N_{i}) = C_{D} (N_{i}) / g - 1

(2)

where C'_D (N_i) denotes the relative degree centrality of the node i, and the denominator denotes g – 1 nodes.

Intermediary mediation is an indicator of information exchange or control of resource flows in the network. It is assumed that node j must communicate with node k through node i, and node i controls the content of information transmitted between node j and node k. The more nodes that appear in the shortest path between pairs of relational nodes ( g and k ), the higher the likelihood that the node i will control network communication. The betweenness centrality calculation method is as shown in Equation (3):

C_{B} (N_{i}) = \sum_{j = 1}^{g} g_{j k} N_{i} / g_{j k}

(3)

where g_jk is the shortest path number between j and k, and g_jkN_i is the shortest path number passing through point i between node j and node k.

Closesness centrality describes the extent to which it is not controlled by any other node. From the point of view of information transmission, the node is closer to other nodes, and this node is easier to transmit information. The calculation method is Equation (4):

C_{C} {(N_{i})}^{- 1} = \sum_{j = 1}^{g} d (N_{i} N_{j})

(4)

where d (N_i N_i) represents the sum of the shortest distances between node i and other g – 1 nodes. The closer the centrality of a node is, the less it is controlled by any other node.

Drawing maps is a relatively common method of displaying network relationships (Gunther & Nagy, 2011). The maps use visualisation and can efficiently reflect and reveal the complex relationships of international collaboration (Morrison et al., 2008). Maps describe the overall characteristics of emerging technology cooperation networks, while centrality analysis is used to measure the importance of nodes in the network structure. Degree centrality is often used as a first step, while betweenness centrality elaborates the ability of a given node to control interactions between pairs of other nodes and closeness centrality describes the degree of a node that is not subject to any other node’s control (de Cannière, 2007).

Global Science Network of Global Hubs

Global Science Cooperation Network of Global Hubs

Before analysing the network characteristics of global hubs, we briefly discuss the evolutionary trend of 15 global hubs in big data analysis and mining technology. Table 1 presents the number of highly cited papers of each global hub from 2013 to 2019. Because of the few papers published from 2000 to 2012, we do not visualise the publications during this period. In Figure 2, the size of the node represents the number of papers published by a particular global hub. Meanwhile, the connection lines reflect the relationship among the global hubs in the previous year and the global hubs in the next year, including strengthening, weakening, generation, extinction, fusion and fission. In terms of basic research about big data analysis and mining technology, it shows that Beijing, Silicon Valley and London have the comparative advantages based on the evolutionary trend in general.

Table 1.

The Number of Research Papers from Global Hubs in Big Data Analysis and Mining Technology.

	2013	2014	2015	2016	2017	2018	2019
Beijing	24	36	94	110	133	139	127
Boston	18	30	43	49	49	48	56
London	13	30	28	68	67	72	54
Hong Kong	8	12	22	30	42	46	40
Sydney	11	17	26	33	33	27	37
Silicon Valley	18	40	58	55	67	62	32
Melbourne	0	10	19	30	24	43	34
Shanghai	0	10	28	25	29	39	33
Singapore	0	12	20	22	17	13	20
Washington	11	16	18	26	22	25	20
New York	7	10	16	26	34	30	24
Seoul	6	9	12	18	25	23	18
Los Angeles	0	20	25	25	24	21	9
Paris	0	9	9	17	10	20	0
Zurich	4	12	0	14	14	16	0

Source: WOSCC.

Figure 2.

Source: WOSCC, processed by ITG.

Subsequently, we conduct the global science cooperation network of global hubs in big data analysis and mining technology. In Figure 3, there are three values in parentheses of each global hub in big data analysis and mining technology. They represent the number of papers published from global hubs in big data analysis and mining technology as the first author, the second author, the third and other authors, respectively. Below the parentheses, there are different subject areas and corresponding papers published by different global hubs. More importantly, the lines between different global hubs reflect the intensity of cooperation in basic research of big data analysis and mining technology. The wider the line, the more the papers are published in the form of co-authors in these two global hubs.

The subject areas that different global hubs attach importance to, can partly explain the intensity of cooperation in basic research of big data analysis and mining technology. It shows that the division of research among different global hubs in big data analysis and mining technology around the world is different. For example, in the field of big data analysis and mining, Beijing pays more attention to the research of computer science and information systems, engineering, electrical and electronic and telecommunications, while Zurich focuses more on Environmental Sciences, Astronomy and Astrophysics and Ecology. The subject areas of the two global hubs in big data analysis and mining technology basically have no overlapping disciplines, so there is very little cooperation between them.

On the other hand, the global science cooperation network shows the characteristic of regionality. As can be seen in Figure 3, Beijing has more cooperation with Shanghai, Hong Kong and Sydney. Meanwhile, Silicon Valley has higher interaction with Los Angel, Boston and London. This shows that despite the international nature of global hubs, they are more inclined to cooperate with domestic cities or neighboring countries. Beijing and Silicon Valley are at the core of the global science cooperation network and play an important role in regional research cooperation. However, there is not much cooperation between these two global hubs in big data analysis and mining technology.

Figure 3.

Source: WOSCC, processed by ITG.

To quantitatively assess the importance of global hubs in the global science cooperation network, we further introduce the centrality indicators. Combining the analysis of centrality indicators (Table 2), we find that Beijing and Silicon Valley have obvious advantages. The two hubs lie in the centricity of the global science cooperation network, which indicates that they are important to the global science cooperation network. Based on the betweenness centrality and the closeness centrality, it shows that Beijing is the most active in global science cooperation network in the field of big data analysis and mining. Greater involvement in the global science cooperation network can largely explain why Beijing can publish the largest amount of highly cited papers in big data analysis and mining technology. However, neither the number of research papers nor the degree of involvement in the global science cooperation network, is necessarily related to the strength and influence of global hubs in basic research, which will be discussed in the next part.

Table 2.

The Centrality Indicators of Global Hubs in Big Data Analysis and Mining Technology.

No.	Global Hubs	Degree Centrality			Betweenness Centrality		Closeness Centrality
No.	Global Hubs	Degree	Nrmdegree	Share	Betweenness	Nbetweenness	Farness	Ncloseness
1	Silicon Valley	375.000	28.111	0.082	2.507	0.617	31.000	93.548
2	Beijing	402.000	30.135	0.088	4.175	1.028	29.000	100.000
3	Boston	290.000	21.739	0.063	3.190	0.786	30.000	96.667
4	London	267.000	20.015	0.058	2.261	0.557	31.000	93.548
5	Washington	209.000	15.667	0.046	2.261	0.557	31.000	93.548
6	Los Angeles	203.000	15.217	0.044	1.647	0.406	32.000	90.625
7	New York	206.000	15.442	0.045	3.898	0.960	30.000	96.667
8	Sydney	162.000	12.144	0.035	1.917	0.472	34.000	85.294
9	Melbourne	166.000	12.444	0.036	4.175	1.028	29.000	100.000
10	Hong Kong	177.000	13.268	0.039	4.175	1.028	29.000	100.000
11	Paris	125.000	9.370	0.027	1.148	0.365	34.000	85.294
12	Shanghai	135.000	10.120	0.029	2.832	0.698	33.000	87.879
13	Singapore	90.000	6.747	0.020	2.058	0.507	33.000	87.879
14	Zurich	74.000	5.547	0.016	0.469	0.115	36.000	80.556
15	Seoul	61.000	4.573	0.013	0.887	0.219	36.000	80.556

Source: WOSCC, processed by Ucinet.

Moreover, Melbourne has strong betweenness centricity in the global science network. Though the degree centricity of Melbourne was not high, maintaining a strong mediation centricity role embodies Melbourne acts as the intermediary hub and bridge role in the global science cooperation network. Hong Kong is really similar to Melbourne. The betweenness centrality is comparatively strong but the degree centrality is not that high. The degree centrality of London, Washington and Los Angeles is strong, but the betweenness centrality is weak. The index of Zurich and Seoul remained the lowest or the second lowest position, had a big disparity with other countries.

Global Science Citation Network of Global Hubs

Global science citation network can effectively reflect the direction of knowledge transfer, then illustrate the influence or power of different global hubs in basic research. If a hub’s papers are cited more frequently by other hubs, it implies that this hub plays a crucial role in global science progress. Figure 4 is global science citation network of global hubs in big data analysis and mining technology. Paris and Zurich are not taken into account because of a limited number of citations so they can not be fully presented in the global science citation network. In this figure, the rectangular labels represent the names of global hubs, and the lines with arrows show the direction of citation. In addition, different width of lines reflects different intensities of citation relationship.

As seen in Figure 4, there is a general citation network among different global hubs in big data analysis and mining technology. On the one hand, at least one-way reference in the technology of big data analysis and mining is presented between two global hubs except for Washington and Singapore. On the other hand, the two-way reference relationship dominates the global science citation network, indicating that no hubs can absolutely monopolise the technology of big data analysis and mining technology, and mutual interaction is mainstream among global hubs in big data analysis and mining technology.

Figure 4.

Source: WOSCC, processed by Ucinet.

However, from the direction and width of the lines, it can be found that there is a distinct performance in the aspect of the influence of global hubs in big data analysis and mining technology. Research papers published in Silicon Valley are cited most by Beijing and London, the top two hubs in terms of the number of published papers in the latest five years. However, there is not as much knowledge flowing from Beijing to Silicon Valley. Additionally, these papers are also widely referenced by other hubs such as Los Angeles, Sydney, Boston, Hong Kong and Singapore. These findings show that Silicon Valley with high-quality research papers can essentially affect the current technology of big data analysis and mining. Besides Silicon Valley, Beijing, London, Boston and Sydney are the top five global hubs in terms of total citations, which implies that these cities have a profound influence in the area of big data analysis and mining.

Last but not least, the global hubs in big data analysis and mining technology also form some ‘small worlds’ in the global science citation network. Based on the citation lines with heterogenous width, it is not hard to find that the interaction in American domestic hubs or Chinese domestic hubs is more common, as well as the hubs along the Atlantic coast or the Pacific coast. In general, the global science citation network has objectively presented the research influence and interactive features in global hubs in big data analysis and mining technology.

Knowledge Management Implication of Global Hubs

The structures and characteristics of global science network in big data analysis and mining technology have an important implication for knowledge management of global hubs. Based on the above visual analysis, we hereby propose a knowledge flow and management model of global hubs (Figure 5). This model fully reflects how knowledge flows among different global hubs and what factors promote or limit the dissemination and absorption of knowledge, which is conducive to improving the efficiency of knowledge creation for global hubs.

Figure 5.

Source: Summarized by authors.

First, knowledge attribute is one of the important factors restricting the global transfer of knowledge. In theory, the citation of academic outcome, as codified knowledge, is not necessarily limited by regional differences, for researchers can share the public knowledge all over the world. Besides, researchers in different global hubs have the access to cooperate in the technology of big data analysis and mining, which can be partially realised by close communication and specialised input through the internet. However, visual analysis of global science network shows that the existence of implicit knowledge generates the phenomenon of regionality or ‘small world’, which may lead to ‘lock-in effect’. In the information society, the tacit, cultural, situational, accidental and other characteristics of implicit knowledge make the exchange of knowledge still unable to get rid of regional restrictions, even among the global hubs in big data analysis and mining technology.

Second, knowledge stock, potential difference of knowledge and compound proximity may affect the knowledge transfer to large extent, then change the network structure among global hubs. We find that the global hubs with more knowledge stock, such as Silicon Valley, Beijing, Boston and London, are more likely to dominate the direction of knowledge transfer. Furthermore, the two global hubs in big data analysis and mining technology with lower potential difference of knowledge tend to cooperate and cite in the aspect of research, for similar knowledge structures and technical systems make the flow of knowledge easier. Compound proximity, including geographical proximity, institutional proximity and discipline proximity is also the key factor to affect the dissemination of knowledge. Empirical results prove that global hubs in big data analysis and mining technology with similar geographical proximity, institutional proximity and discipline proximity have more connection during the research. All these dimensions change the knowledge flow among global hubs in big data analysis and mining technology, and finally shape the global science network.

Third, global hubs can be divided into central hubs and edge hubs in the light of network scale, network centrality and network connection strength. For example, Silicon Valley, Beijing, Boston and London can be classified into central hubs in big data analysis and mining technology, with network influence and resource integration ability. Visual outcome reflects that knowledge spillover and preferential attachment between two kinds of hubs in global science network are more common, like Beijing with Seoul, Hongkong and Sydney. During this process, specialised division of labor, shared production factors and diversified market environment are the driving force to detect new technology supply and demand, which could contribute to knowledge creation. Therefore, the existence of hierarchical relationship in global science network is necessary for science progress of global hubs in big data analysis and mining technology.

To sum up, strengthening the whole process management of knowledge stock, knowledge transfer, knowledge network and knowledge creation, and promoting the virtuous cycle of knowledge flow, are important factors for global hubs in big data analysis and mining technology to take the lead in particular frontier technology.

Conclusions and Discussions

Conclusions

To explore the global science network of 15 representative global hubs, we adapted the method of social network analysis and selected highly cited papers in the area of big data analysis and mining technology from 2000 to 2019. Based on the structures of global science citation network and characteristics of global science cooperation network, the knowledge flow and management model of global hubs was proposed and several conclusions have been drawn as follows.

First, comprehensive evaluation criteria should be established to evaluate the science position of global hubs. From evolutionary trend and global science cooperation network, Beijing seems to have the leading advantage in basic research of big data analysis and mining technology. However, it should not be too hasty to conclude that Beijing is dominant in global hubs in big data analysis and mining technology. Based on the global science citation network, research papers from Silicon Valley are cited mostly by Beijing and London, while there is not as much knowledge flowing from Beijing to Silicon Valley. Hence, compared with the quantity advantage of research in Beijing, the strengths of basic research in Silicon Valley lie in originality and quality. In the future, more comprehensive evaluation standards should be taken into account to evaluate the science position of global hubs objectively.

Second, the global science network has the obvious characteristic of regionality, which does not coincide with its global property. Hubs in the same country (Silicon Valley, Los Angeles and Boston in U.S., Beijing, Shanghai and Hong Kong in China) or in a close area (along the Atlantic coast and the Pacific coast) are more likely to research jointly and cite mutually. This finding is not consistent with some previous research (Ernst, 2008), which believes that geographical distance is no longer the main obstacle to the formation of a global science network in this information society. On one hand, the tacit, cultural, situational, accidental and other characteristics of implicit knowledge make the exchange of knowledge still unable to get rid of regional restrictions. On the other hand, visual analysis shows that global hubs in big data analysis and mining technology with similar geographical proximity, institutional proximity and discipline proximity have more connections during the basic research.

Third, the global science network has the structure of hierarchy, which is conducive to accelerating the flow of scientific elements. Global hubs can be divided into central hubs and edge hubs in the light of network scale, network centrality and network connection strength. Visual outcome reflects that knowledge spillover and preferential attachment between two kinds of hubs in global science networks are more common, like Beijing with Seoul, Hongkong and Sydney. During this process, a specialised division of labor, shared production factors and diversified market environment are the driving force to detect new technology supply and demand, which could contribute to knowledge creation. Therefore, the existence of a hierarchical relationship in the global science network is necessary for science progress of global hubs in big data analysis and mining technology.

Discussions

However, wo main limitations can be discussed in future research. First, we extract the address information of the authors’ institutions and build the global science cooperation network and global science citation network. Due to some constraints, we do not classify the attributes of authors’ institutions, such as universities or large enterprises (like Amazon and Microsoft). In the future, the interaction among the main actors of global hubs in the global science network, especially universities and enterprises as carriers of knowledge flow and creation, can be further discussed. That is conducive to exploring the micro-mechanism of the network structures and characteristics of global hubs.

Second, to avoid the interference of exogenous factors, this paper avoids the discussion about the performance of global science network after COVID-19, international politics and trade conflict. Without these exogenous factors, we successfully propose a knowledge flow and management model of global hubs in general. However, with the increasingly complex external situation, the global science network will face more uncertainties in the future. How to deal with discontinuous external shocks in global science network is a major challenge for global hubs, which needs to be further explored to strengthen the flow and management of knowledge.

Footnotes

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.

Funding

This work is supported by the Chinese National Funding of Social Sciences (No. 20&ZD075).

Notes>

References

Arif

, Alghamdi

T. A.

, Khan

Z. A.

, & Javaid

(2022). Towards efficient energy utilisation using big data analytics in smart cities for electricity theft detection. Big Data Research, 27, 100285. 10.1016/j.bdr.2021.100285

Bachechi

, Po

, & Rollo

(2022). Big data analytics and visualisation in traffic monitoring. Big Data Research, 27, 100292. 10.1016/j.bdr.2021.100292

Balduini

, Brambilla

, Della Valle

, Marazzi

, Arabghalizi

, Rahdari

, & Vescovi

(2019). Models and practices in urban data science at scale. Big Data Research, 17, 66–84. 10.1016/j.bdr.2018.04.003

Bai

, Wu

, Liu

, & Xu

(2020). Research on the impact of global innovation network on 3D printing industry performance. Scientometrics, 124(2), 1015–1051. 10.1007/s11192-020-03534-1

Busher

H. H.

(1976). Basic scientific research. Chest, 70(4), 500–500. 10.1016/S0012-3692(16)39281-9

Chaminade

, & Plechero

(2015). Do regions make a difference? Regional innovation systems and global innovation networks in the ICT industry. European Planning Studies, 23(2), 215–237. 10.1080/09654313.2013.861806

Clark

, Harrison

, & Miguelez

(2018). Connecting cities, revitalising regions: The centrality of cities to regional development. Regional Studies, 52(8), 1025–1028. 10.1080/00343404.2018.1453691

de Beaver

D. B.

, & Rosen

(1979). Studies in scientific collaboration: Part II—scientic co-authorship, research productivity and visibility in the French scientific elite, 1799–1830. Scientometrics, 1(2), 133–149. 10.1007/BF02016966

de Cannière

(2007). The seven stages of an idea. Journal of Thoracic and Cardiovascular Surgery, 134(3), 562–564. 10.1016/j.jtcvs.2007.03.035

10.

Debin

, & Shunhui

(2016). Connotation, function and organisational structure of global science and Technology Innovation Center. Forum on Science and Technology in China, 25(2), 10–15.

11.

Delu

(2014). Science and Technology Innovation Center from the Perspective of New Economy. Zhongguancun, 13(6), 73–75.

12.

Derudder

, & Taylor

P. J.

(2018). Central flow theory: Comparative connectivities in the world-city network. Regional Studies, 52(8), 1029–1040. 10.1080/00343404.2017.1330538

13.

Ernst

(2008). Innovation offshoring and Asia’s electronics industry—The new dynamics of global networks. International Journal of Technological Learning, Innovation and Development, 1(4), 551–576.

14.

Freeman

(1991). Networks of innovators: A synthesis of research issues. Research Policy, 20(5), 499–514. 10.1016/0048-7333(91)90072-X

15.

Guan

, Zhang

, & Yan

(2015). The impact of multilevel networks on innovation. Research Policy, 44(3), 545–559. 10.1016/j.respol.2014.12.007

16.

Gunther

F. H.

, & Nagy

(2011). Global program for Collaborative Research, Development and Innovation: Introduction to IMS international. IEEE International Conference Mechatronics and Automation, Beijing, China. 10.1109/ICMA.2011.5985622

17.

Human Development Report. (2001). United Nations Development Program (UNDP): Making new technologies serve human development.

18.

Imran

, Mahmood

, Morshed

, & Sellis

(2021). Big data analytics in healthcare A systematic literature review and roadmap for practical implementation. IEEE/CAA Journal of Automatica Sinica, 8(1), 1–22. 10.1109/JAS.2020.1003384

19.

Jian

, & Qiyu

(2016). Geographical pattern of international urban innovation competitiveness from the perspective of global innovation network. Social Sciences, 22(9), 25–38.

20.

Jianwen

, Haijun

, & Hong

(2015). Construction experience of international city group Technology Innovation Center and its enlightenment to Shanghai. East China Science Technology, 12(3), 64–68.

21.

Junjie

, Mengke

, & Rui

(2019). Research on perfecting the land use system to promote the construction of Shanghai global science and Technology Innovation Center. Science and Technology and Economy, 32(02), 35–39.

22.

Katz

J. S.

, & Martin

B. R.

(1997). What is research collaboration? Research Policy, 26(1), 1–18. 10.1016/S0048-7333(96)00917-1

23.

Krishna

V. V.

(2017). Inclusive innovation in India: Contemporary landscape. Asian Journal of Innovation and Policy, 6(1), 1–22. 10.7545/ajip.2017.6.1.001

24.

Krishna

V. V.

(2022). India @ 75: Science, technology and innovation policies for development. Science, Technology and Society, 27(1), 113–146. 10.1177/09717218211047029

25.

Lee

(2017). Big data: Dimensions, evolution, impacts, and challenges. Business Horizons, 60(3), 293–303. 10.1016/j.bushor.2017.01.004

26.

, Yang

, Qu

, & Kong

(2022). Research on the formation mechanism of big data technology cooperation networks: Empirical evidence from China. Scientometrics, 127, 1273–1294. 10.1007/s11192-022-04270-4

27.

Marcus

S. E.

, Moy

, & Coffman

(2006). Social network analysis. Mining graph data. John Wiley & Sons, Inc.

28.

Matthiessen

C. W.

, Schwarz

A. W.

, & Find

(2001). The top-level global research system, 1997–99: Centers, networks and nodality—An analysis based on bibliometric indicators. Urban Studies, 39(5/6), 903–927

29.

Morrison

, Pietrobelli

, & Rabellotti

(2008). Global value chains and technological capabilities: A framework to study learning and innovation in developing countries. Oxford Development Studies, 36(1), 39–58. 10.1080/13600810701848144

30.

Ran

& Liu

(2023). The change and continuity in China’s S&T policy: An institutional grammar (IG) approach. Asian Journal of Technology Innovation. 10.1080/19761597.2023.2181832

31.

Scott

A. J.

(Ed.). (2001). Global city-regions: Trends, theory, policy. Oxford University Press.

32.

Watson-Fargie

, & Cheape

(2018). Ensuring quality immediate discharge letters from mental health. International Journal of Health Care Quality Assurance, 31(6), 545–551.

33.

Wolfe

A. W.

(1995). Social network analysis: Methods and applications, by Stanley Wasserman; Katherine Faust. Contemporary Sociology, 91(435), 219–220.

34.

Xiaozan

, Xiaojing

, & Ping

(2021). Visual analysis on global big data collaboration network under the background of science and technology globalisation. Science and Technology Management Research, 41(16), 26–36.

35.

Ying

, Zihan

, Jing

, & Qi

(2015). Bibliometric analysis of Big Data research progress. Chinese Science and Technology Information, 527(24), 79–83.

36.

Yun

J. J.

, Jeong

, & Yang

J. H.

(2015). Open innovation of knowledge cities. Journal of Open Innovation: Technology, Market, and Complexity, 1(1), 16. 10.1186/s40852-015-0020-x

37.

Yun

J. H. J.

(2022). About capital and ideology by Tomas Piketty. Journal of Open Innovation: Technology, Market, and Complexity, 8(26), 1–15.

38.

Yun

J. J.

, Park

K. B.

, Hahm

S. D.

, Kim

(2019). Basic Income with high Open Innovation Dynamics: The way to the Entrepreneurial State.. Journal of Open Innovation: Technology, Market, and Complexity, 5(3), 41.

39.

Yuqin

, Xuefeng

, & Xiaoping

(2015). Design and implementation of scientific research relationship construction and visualisation system. Library and Information Service, 8, 9.