Evolution and Mechanism of Cooperative Innovation Networks Based on Strategic Emerging Industries: Evidence from Northeast China

Abstract

The rapid flow of resources and participation of diverse entities in the information age drive the formation of innovation networks, yet how these networks evolve in strategic emerging industries—particularly in transitioning regions like Northeast China—remains underexplored. This study investigates the spatial formation and evolution of cooperative innovation networks in strategic emerging industries, as well as the underlying mechanisms driving their development and transformation. Focusing on China’s three northeastern provinces—a representative old industrial base undergoing transformation—we construct and analyze inter- and intra-city cooperative innovation networks from 2009 to 2023 using invention patent data. Employing social network analysis and spatial econometric models, we reveal that while the external network exhibits limited expansion with Beijing as a dominant hub, the internal network demonstrates increasing cohesion, transitioning to a multi-core structure. Factors such as urban innovation capacity, economic proximity, and institutional proximity play key roles in shaping these networks. These findings offer policymakers empirical evidence to strategically optimize cross-regional innovation ecosystems in resource-dependent regions, supporting urban sustainability and fostering innovation-driven development.

1. Introduction

Innovation is widely recognized as the primary driver of economic development, particularly within the domain of strategic emerging industries. These industries, defined by their high-tech nature, substantial growth potential, and technological advantages [1], include sectors such as information technology, new energy, and biomedicine. Their strategic importance lies in their ability to drive global economic transformation, enhance industrial upgrading, and strengthen national competitiveness [2]. The primary essence of strategic emerging industries lies in their strategic significance, which is reflected in several aspects. These industries are typically green, low-carbon, and environmentally friendly, aligning with the long-term and sustainable development goals of cities. Moreover, they are closely integrated with government strategies for urban development. The technologies employed in these industries often possess strategic breakthrough potential, relying on innovation and technology-driven development.

However, to fully realize this potential, the achievement of high-quality development in these industries depends on innovation networks based on collaborative relationships. These networks, originally proposed by Freeman [3], facilitate technological synergy and resource sharing among firms, academic and research institutions, and government agencies, thereby accelerating technological breakthroughs and industrialization.

With the advancement of strategic emerging industries, the development of innovation activities increasingly depends on inter-organizational and cross-regional network collaboration. The transition from “local space” to “mobility space” and “cyberspace” has significantly reshaped the dynamics of innovation networks [4,5], driving the reconfiguration of city-centered innovation within cyberspace [6]. Globalization, informatization, and networking have facilitated the strengthening of inter-city connections, disrupting the traditional hierarchical urban organization model and giving rise to a networked urban structure. The innovative capacity of cities is not only dependent on internal innovation resources but is also reflected in the increased connectivity of external networks. These synergistic networks enhance the complementarity and advancement of technology and knowledge [7].

To construct an effective city innovation network, it is essential to quantify the innovation linkages between cities. However, the cooperation process involves a significant amount of tacit knowledge transfer and diffusion, which is difficult to measure directly. Consequently, scholars often use joint patent and collaborative research paper data to construct technology innovation networks and scientific knowledge networks, respectively. Patent collaboration data reflect the cooperation between cities in technological development [8], while co-authored papers between research institutions indicate the distribution and flow of academic resources [9]. These data provide a foundation for quantifying the structural characteristics and evolution of urban innovation networks.

Research on innovation networks, both domestically and internationally, primarily focuses on three key areas. Firstly, case studies of different types of strategic emerging industries are conducted. Domestic research predominantly centers on innovation networks in sectors such as electronic information [10] and biomedicine [11], investigating their formation mechanisms and characteristics. In contrast, international scholars tend to focus more on areas such as artificial intelligence [12,13] and biotechnology [14], examining cross-border collaboration, technology diffusion, and the flow of innovation resources within the global value chain. However, most of these studies are industry-specific, making it challenging to comprehensively capture the characteristics of innovation networks in the context of regional cooperation. As a critical region for advancing strategic emerging industries and facilitating economic restructuring in China, the three northeastern provinces possess unique regional characteristics and development needs in the construction of inter-city and trans-regional innovation networks. Consequently, their innovation network development patterns and evolutionary mechanisms warrant in-depth exploration.

Secondly, research on the spatio-temporal evolution of innovation networks in strategic emerging industries has gained significant attention. The evolution of innovation networks has become a key topic for scholars in economic geography [15], primarily focusing on the changes in topological structure, spatial patterns, and linkages among innovation actors [16]. However, the evolution of innovation networks within strategic emerging industries, particularly the spatial characteristics of inter-city and cross-regional networks, remains an area requiring further investigation.

Thirdly, research on the factors influencing the formation and evolution of innovation networks has focused on both internal and external determinants. Internal factors include the composition of innovation agents and their collaborative linkages, while external factors involve the innovation environment and policy frameworks. In recent years, scholars, notably Boschma [17], have investigated how various proximities, such as geographical, cognitive, and organizational proximities, facilitate or restrict knowledge flow and innovation diffusion, influencing the construction and evolution of innovation networks [18]. The interaction of these multidimensional proximities has been identified as a critical factor in fostering innovation cooperation. However, the impact of proximity varies across different stages of development, spatial scales, and industrial contexts [19]. Further research is needed to understand how different proximities shape the evolution of innovation networks and cooperation in strategic emerging industries, particularly in the three northeastern provinces.

In terms of research methods for innovation networks, Jaffe introduced social network analysis (SNA), which has become the primary method for studying inter-regional innovation linkages [20]. SNA is used to examine the characteristics, topological relationships, and structural evolution of innovation networks. Regarding influencing factors, commonly employed methods include the negative binomial regression model [21] and QAP regression analysis [22], which are used to explore the endogeneity of urban innovation capacity and its spatial spillover effects.

Based on the above discussion, the key issues this study aims to address are as follows: (1) What are the spatial and structural formations and evolutions of the cross-regional cooperative innovation networks in strategic emerging industries in the three northeastern provinces? (2) What are the spatial and structural formations and evolutions of the inter-city cooperative innovation networks in strategic emerging industries in the three northeastern provinces? (3) What role does multidimensional proximity play in shaping the evolution of these networks?

This study examines the spatial and structural evolution of innovation networks in strategic emerging industries across China’s three northeastern provinces, leveraging patent data from 2009 to 2023. By employing ArcGIS10.8 and social network analysis (SNA), this research traces the development of cross-regional and inter-city innovation networks. Additionally, a negative binomial gravity regression model is applied to investigate the underlying mechanisms of network evolution through the lens of multidimensional proximity. This study aims to systematically examine how innovation strategies can drive sustainable industrial transformation in Northeast China, with a focus on optimizing cross-regional and inter-city innovation networks to inform evidence-based policies for revitalizing the region’s economy through technology-led development pathways.

The main contributions of this study are as follows: (1) Focusing on the three traditional industrial bases in Northeast China, this study provides an in-depth analysis of the evolution mechanism and driving factors of urban innovation networks within the context of the region’s ongoing economic transformation. This region, once heavily reliant on traditional industries, is undergoing a profound shift toward strategic emerging industries. Understanding the mechanisms behind innovation network evolution in this context is critical for policymakers, as it offers insights into how these regions can navigate their transition from industrial decline to economic revitalization. This study addresses a significant gap in the literature by providing an empirical model specifically tailored to the needs of declining industrial regions, an area that has received limited attention in existing research. (2) This study broadens the scope by examining the entire spectrum of strategic emerging industries, rather than focusing on single industries. It explores the role of cross-industry synergies in the evolution of regional innovation networks, addressing a significant gap in previous studies that have overlooked inter-industry connections. (3) This study introduces a novel “inter-regional and intra-city” framework, analyzing both external collaborations and internal synergies within urban innovation networks. This multi-scale approach provides new insights into how innovation-driven development unfolds in transitioning regions, offering a more comprehensive view compared to traditional studies focused solely on intra-regional interactions.

2. Material and Methods

2.1. Study Area

The three northeastern provinces of China—Liaoning, Jilin, and Heilongjiang—encompassing 36 prefecture-level administrative units (see Figure 1), are critical old industrial bases and hold a key position in the national regional innovation strategy. With a combined resident population of 85.1241 million in 2023 (6.04% of China’s total), this region forms a cohesive geographic and socioeconomic entity characterized by shared historical trajectories, industrial structures, and policy frameworks. As the core target of China’s “Northeast Revitalization” strategic initiatives since 2003, these provinces face systemic challenges distinct from other regions. However, on a national scale, the northeast region lags behind economically developed areas, such as Beijing–Tianjin–Hebei, the Yangtze River Delta, and the Pearl River Delta, particularly in terms of economic development and scientific and technological innovation capacity. Consequently, there is a significant gap in the development of strategic emerging industries [23,24]. This region faces challenges in industrial upgrading and innovation-driven development, but at the same time, it offers significant opportunities for exploring new models of innovation cooperation.

The innovation environment in the three northeastern provinces exhibits distinct regional characteristics. On one hand, state-owned enterprises (SOEs) dominate the innovation landscape, with a well-established industrial system but limited innovation vitality. On the other hand, the secondary industry accounts for a high proportion of the economic structure, with traditional manufacturing sectors largely driving regional economic development, while high-tech industries remain relatively underdeveloped [25]. Moreover, compared to the southern coastal regions, the northeast has a lower degree of marketization, a relatively rigid institutional environment, and a less developed private economy. In terms of inter-city and cross-regional cooperation, the model in this region relies heavily on policy support, with relatively weak market-driven cooperation dynamics [26]. These factors contribute to an uneven distribution of internal resources and limited synergistic efficiency within the innovation networks, reflecting certain imbalances. As such, the three northeastern provinces provide a typical and representative case for study.

2.2. Data Sources and Data Processing

Patents, as a product of technological innovation, serve as an ideal indicator of innovation cooperation between cities, with cooperative patents representing their primary form [27]. The data on cooperative patents are sourced from the Wisdom Sprout (PatSnap) Global Patent Database (https://www.zhihuiya.com/, accessed on 28 May 2024), utilizing the SEIC classification of the 2018 version of Strategic Emerging Industries to identify nine industry categories. The patent right holders considered in this study include companies, institutions, and government entities. Since China officially introduced the “strategic emerging industries” strategy in 2009, the research period for this study spans from 2009 to 2023, with a total of 389,355 authorized invention patents being screened for analysis.

Pre-processing of patent data: (1) Data Screening: Patents filed individually were removed, and those involving individuals (where addresses could not be confirmed), as well as patents from Hong Kong, Macao, Taiwan, and foreign enterprises, were excluded. This ensures that this study focuses solely on the cooperative innovation network within the three northeastern provinces, excluding any transnational external influences. As a result, 26,913 valid collaborative patents were retained. (2) Multiple cooperation records: Patents with three or more co-applicants were recorded as multiple cooperation records using pairwise combinations. (3) Geographic information matching: Using the “Tianyancha” website and manual searching, each applicant was matched with their respective city. (4) Constructing an undirected weighted matrix: Based on the joint applications of invention patents between cities, an undirected weighted matrix was established with cities as nodes, patent cooperation relationships as edges, and the number of cooperations as weights. This matrix was used to construct a city cooperation and innovation network in the three northeastern provinces.

This study divides the research period into three time intervals—2009–2013, 2014–2018, and 2019–2023—using a five-year boundary to systematically analyze the evolution of the urban innovation network for strategic emerging industries in the three northeastern provinces.

2.3. Method and Model Selection

2.3.1. Social Network Analysis

Social network analysis is a crucial method for studying the interrelationships within networks, revealing both the structural characteristics of the overall network and individual nodes [11]. It is widely used in innovation network studies. We selected network size, connection strength, and small-world indicators to examine the overall characteristics of the cooperative innovation network among northeastern cities. Additionally, the positions of city nodes in the network were assessed based on their centrality.

2.3.2. Geographical Information System (GIS) Spatial Visualization Method

In the spatial analysis, cooperation between innovation entities was converted into cooperation between cities based on the locations of the applicants. The XY to Line function in GIS was then used to derive the spatial pattern map of the innovation network [28].

2.3.3. Negative Binomial Gravity Model

The gravity model theory posits that interactions between entities are primarily influenced by the entities’ own attributes and the spatial distance between them. Scherngell [29] proposed a negative binomial regression method based on the gravity model to analyze the mechanisms via which various factors impact inter-regional innovation cooperation [30]. In this study, we adopted the model proposed by these researchers and used it to examine the influencing factors of the internal innovation network in the three northeastern provinces across the three time periods and to analyze its evolutionary mechanisms. The specific model was set up as follows:

$$\begin{array}{ll}{C}_{ij}& =\alpha +{\beta }_{1}PA{T}_i+{\beta }_{2}PA{T}_j+{\beta }_{3}Ge{o}_{ij}+{\beta }_{4}Bo{r}_{ij}+{\beta }_{5}Ec{o}_{ij}+{\beta }_{6}In{d}_{ij}\\ & +{\beta }_{7}In{s}_{ij}+{\beta }_{8}Te{c}_{ij}+{\beta }_{9}So{c}_{ij}\cdot \mathrm{gov}+{\beta }_{10}So{c}_{ij}\cdot \mathrm{edu}+{\epsilon }_{\mathit{ij}}\end{array}$$

(1)

where $C_{ij}$ is the intensity of cooperation between city i and city j; $\alpha$ is the constant term; $\beta$ is the estimated parameter; and ${\epsilon }_{ij}$ is the random error term. The meanings and data sources of the variables are detailed in

Table 1. Explanations of the independent variables in the negative binomial gravity regression model.

Variable	Explanation	Data Sources and Processing
PATi	Total number of patent applications for inventions in city i	The number of patent applications for each time period is a five-year average.
PATj	Total number of patent applications for inventions in city j	The number of patent applications for each time period is a five-year average.
Geoij	Normalized spatial distance between city i and city j. Together with Borij, this characterizes the geographic proximity between cities.	$Ge{o}_{ij}=1-\ln \left({\displaystyle \frac{d_{ij}}{\max d_{ij}}}\right)$	(2)
		dij denotes the geographic distance between node i and node j, calculated by city latitude and longitude, and maxdij is the maximum distance in the sample.
Borij	Administrative boundary proximity between city i and city j. Together with Geoij, this characterizes the geographic proximity between cities.	A dummy variable that takes the value of 1 if the two cities share an administrative boundary
Ecoij	Degree of similarity in the level of economic development between city i and city j. This characterizes the economic proximity between cities.	Expressed as inter-city GDP ratio.
Indij	Degree of similarity in the level of industrial structure between city i and city j. This characterizes the industrial proximity between cities.	Expressed as the ratio of tertiary value added to GDP between cities [31].
Insij	Degree of institutional difference between city i and city j. This characterizes institutional proximity between cities.	Cities in the same province are assigned a value of 1, while different provinces are assigned 0.
Tecij	Degree of similarity in technological knowledge between city i and city j. This characterizes technological proximity between cities.	${Tec}_{ij}={\displaystyle \frac{{\displaystyle {\sum }_{m=1}^{8}PA{T}_{imt}\times PA{T}_{jmt}}}{\sqrt{{{{\displaystyle {\sum }_{m=1}^{8}PAT}}^2}_{\mathrm{imt}}\times PA{T^2}_{\mathrm{jmt}}}}}$	(3)
		Drawing on Jaffe’s technological proximity measure formula [32], PATimt and PATjmt are the number of patent applications for cities i and j in year t in category m-th IPC classification number. The values range from 0 to 1, with larger values indicating greater technological proximity. The number of patents for each period is calculated using the sum of patent numbers over five-year intervals.
Socij·gov	The degree of similarity in governmental support between city i and city j, with Socij·edu characterizing social proximity between cities.	Expressed as the ratio of financial expenditures on science and technology between cities.
Socij·edu	The extent to which city i and city j have similar levels of education, with Socij·gov characterizing social proximity between cities.	Expressed as the ratio of financial expenditures on education between cities.

Note: (1) Eco_ij, Ind_ij, Soc_ij·gov, and Soc_ij·edu are calculated as the ratios between the two cities, with the maximum value as the denominator and the minimum value as the numerator, resulting in a range of (0, 1). (2) The data for PAT_i, PAT_j, and Tec_ij were sourced from the Shanghai Intellectual Property Information Service Platform. (3) Eco_ij, Ind_ij, Soc_ij·gov, and Soc_ij·edu data were obtained from the “China City Statistical Yearbook” and city statistical bulletins, with a five-year average taken for each period. (4) Yanbian Korean Autonomous Prefecture and Daxing’anling Prefecture were excluded from the analysis due to data unavailability.

, details of the specific data are provided in the Supplementary Material.

3. Results

3.1. Evolution Characteristics of the External Inter-City Cooperative Innovation Network in the Three Northeastern Provinces

3.1.1. Evolution Characteristics of the Topological Structure of the External Inter-City Innovation Network

(1) Network expansion and loose density. With the development of strategic emerging industries in the three northeastern provinces, external connections have become increasingly close. The external innovation network in the three northeastern provinces has experienced significant growth over the years, as indicated by the increase in both the number of participating cities and collaborative ties (see

Table 2. Overall characteristics of external cooperative innovation networks in the three northeastern provinces.

	Indicator	2009–2013	2014–2018	2019–2023
Network size	Number of network nodes	149	186	245
	Number of network relationships	293	453	775
	Network density	0.02657	0.02633	0.02593
Contact strength	Average degree	3.933	4.871	6.327
	Average weighted degree	34.161	68.022	125.478
Small-world characteristics	Average path length	2.797	2.778	2.669
	Average clustering coefficient	0	0	0.002

). Specifically, the node count surged by 64% (149 cities in 2009–2013→245 cities in 2019–2023), while the number of collaborative ties expanded by 165% (293→775 relationships). However, despite this expansion, the network density declined slightly by 2.4% (from 0.02657 to 0.02593). This paradox, characterized by “growth without densification”, reflects a limited integration of newly added cities into the core hubs of Northeast China. Consequently, the external network has developed an “umbrella-like” structure, where newly joined cities contribute only sporadically to the overall connectivity, leading to a dilution in network density.

(2) Increasingly close innovation ties, lacking small-world characteristics. As shown in Table 2, the average weighted degree centrality has increased by 267% (from 34.161 to 125.478), reflecting a deepening of bilateral collaborations across cities. Small-world characteristics typically manifest as a high clustering coefficient (tight local connections) and a low characteristic path length (efficient global connections) [33]. These features allow nodes in small-world networks to reach other nodes with few steps, although they are relatively rarely directly connected. Generally, a network is considered to have small-world properties if the clustering coefficient is greater than 0.1 and the average path length is less than 10 [34]. However, the clustering coefficient remains near zero (<0.002), which is well below the small-world threshold of 0.1, indicating that local clusters are weak and fail to form dense, tightly knit groups. The average path length has decreased slightly (from 2.797 to 2.669), suggesting improved regional connectivity. These observations suggest that the network has expanded broadly but lacks the high level of local interconnectedness characteristic of small-world networks, which would typically facilitate more efficient information flow and collaboration.

(3) External cooperation dominated by provincial capitals and municipalities, with evident spatial polarization. Beijing maintained its position as the dominant super-core hub throughout the entire period from 2009 to 2023 (

Table 3. Top-10 cities by centrality in the Northeast China-nationwide inter-city cooperative innovation network.

Indicator	Degree Centrality			Weighted Degree Centrality			Betweenness Centrality
Ranking	2009–2013	2014–2018	2019–2023	2009–2013	2014–2018	2019–2023	2009–2013	2014–2018	2019–2023
1	Beijing *	Beijing *	Beijing *	Beijing *	Beijing *	Beijing *	Beijing *	Beijing *	Beijing *
2	Shanghai *	Tianjin *	Shanghai *	Shanghai *	Shanghai *	Nanjing *	Shanghai *	Nanjing *	Xi’an *
3	Tianjin *	Hangzhou *	Tianjin *	Yiyang	Nanjing *	Shenzhen	Tianjin *	Shanghai *	Tianjin *
4	Nanjing *	Nanjing *	Xi’an *	Nanjing *	Shenzhen	Shanghai *	Nanjing *	Tianjin *	Shanghai *
5	Wuxi	Shanghai *	Nanjing *	Tianjin *	Tianjin *	Xi’an *	Xi’an *	Hangzhou *	Shenzhen
6	Shenzhen	Shenzhen	Shenzhen	Suzhou	Qingdao	Ningbo	Wuxi	Shenzhen	Nanjing *
7	Chengdu *	Suzhou	Wuhan *	Shenzhen	Xuchang	Chongqing *	Wuhan *	Suzhou	Wuhan *
8	Qingdao	Shijiazhuang *	Hangzhou *	Chengdu *	Suzhou	Tianjin *	Guangzhou *	Shijiazhuang *	Hangzhou *
9	Xi’an *	Kunming *	Qingdao	Qingdao	Wuhan *	Hangzhou *	Chengdu *	Kunming *	Qingdao
10	Kunming *	Chengdu *	Guangzhou *	Guangzhou *	Guangzhou *	Chengdu *	Shenzhen	Urumqi *	Guangzhou *

* Provincial capitals and municipal cities.

). It consistently ranked first across all centrality metrics—degree centrality, weighted degree centrality, and betweenness centrality—underscoring its central role in the innovation network. Additionally, the influence of provincial capitals and municipalities has grown over time, with these cities comprising 80% of the top-10-ranked cities by 2019–2023, compared to 60% in 2009–2013. Among these emerging subcores, Nanjing saw a significant rise in weighted centrality, advancing from 4th to 1st place by 2019–2023, while Shenzhen also improved its degree centrality rank, moving from 6th to 5th. In contrast, non-administrative hubs like Suzhou gradually lost their prominence, dropping out of the top-10 rankings by the end of the study period. This trend highlights the increasing dominance of administrative centers in the network, reinforcing a core–periphery structure, in which cities like Beijing continue to anchor the network, while non-administrative cities in regions like the Beijing–Tianjin–Hebei, Yangtze River Delta, and Greater Bay Area struggle to maintain their position.

3.1.2. The Spatial Evolution Characteristics of External Inter-City Innovation Network Structures

(1) The spatial distribution has gradually become more concentrated over the years, with significant clustering in the eastern and southern regions. Cities in the three northeastern provinces have increasingly engaged in innovation cooperation with cities nationwide, and the spatial distribution of external cooperation cities has expanded each year. From 2009 to 2013, the primary clusters were concentrated in the Beijing–Tianjin–Hebei region and the Yangtze River Delta. From 2014 to 2018, the clustering extended further south, focusing on the Pearl River Delta. Between 2019 and 2023, the distribution expanded toward the Chengdu–Chongqing city cluster and the Wuhan metropolitan area in the central and western regions. Many cities in the central and western regions have ended their isolation, becoming new members of the innovation network. The network structure has evolved toward greater complexity, ultimately displaying a pattern of “dense in the southeast, sparse in the northwest” (see Figure 2).

(2) In terms of network cooperation intensity, Beijing stands out as a prominent engine of innovation, with the influence of municipalities, provincial capitals, and sub-provincial cities steadily increasing. Most central nodes and key cooperative relationships are concentrated on the eastern side of the “Hu Huanyong Line” [35]. Using the natural breaks method, external innovation cooperation links are categorized into three levels: high intensity (frequency [501, 8664]), medium intensity (frequency [26, 500]), and low intensity (frequency [1, 25]). In the high-intensity category, the proportion of cooperation involving Beijing increased from 61.8% to 65.27% before decreasing to 57.61%. Despite the decline in cooperation intensity later on, the polarization effect remains evident. Medium-intensity cooperative cities are concentrated among municipalities, provincial capitals, and sub-provincial cities, such as Shanghai, Nanjing, and Shenzhen. These cities are gradually becoming secondary hubs, with their cooperation intensity and network dominance increasing considerably. Low-intensity cooperation is widespread across the country but generally weakens, especially in regions such as Tibet, Qinghai, and Gansu, where participation is limited, and the cooperation network is fragile.

3.2. The Evolution Characteristics of the Inter-City Cooperative Innovation Network Within the Three Northeastern Provinces

3.2.1. Evolution Characteristics of the Topological Structure of the Inter-City Cooperative Innovation Network

(1) The network scale has gradually expanded and become more compact. From 2009 to 2013, the city cooperation rate was 83%, with the Heihe, Yichun, Qitaihe, Liaoyuan, and Daxing’anling area not participating (

Table 4. Overall characteristics of intra-regional cooperative innovation networks.

	Indicator	2009–2013	2014–2018	2019–2023
Network size	Number of network nodes	30	34	36
	Number of network relationships	74	107	166
	Network density	0.17011	0.19073	0.26349
Contact strength	Average degree	4.933	6.294	9.222
	Average weighted degree	40.067	65.235	198.611
Small-world characteristics	Average path length	2.103	1.95	1.768
	Average clustering coefficient	0.618	0.669	0.66

). Between 2014 and 2018, the cooperative innovation network continued to deepen, with only the Yichun and Daxing’anling area remaining isolated. From 2019 to 2023, all cities in the three northeastern provinces engaged in cooperative innovation, with increasingly frequent connections, witnessing a 20% growth in participating cities (30→36) and a 124% surge in collaborative ties (74→166), accompanied by a 55% increase in density, from 0.17011 to 0.26349. In the topological structure diagram (see Figure 3), Shenyang, Dalian, Changchun, and Harbin emerged as the main cooperative cities. Shenyang, Dalian, Changchun, and Harbin emerged as the main cooperative cities. This reflects enhanced peripheral integration through anchor hubs like the Shenyang–Dalian–Changchun–Harbin cluster, which mediated 68% of new connections post-2014. The intra-regional cooperative innovation network in Northeast China demonstrated marked structural transformations between 2009 and 2023, characterized by synergistic scaling, emergent small-world efficiency, and dynamic core–periphery reconfiguration.

(2) The intensity of cooperation has increased, exhibiting small-world characteristics. The clustering coefficient, a critical indicator of small-world properties, remained above 0.618, which is over six-times the typical threshold of 0.1 for small-world networks. This reflects the emergence of tight-knit local clusters within the network. Additionally, the average path length decreased by 16%, from 2.103 to 1.768, crossing the small-world efficiency threshold of <2.0, which suggests a more efficient global connection within the network. Furthermore, the average weighted degree, which measures the intensity of connections, increased by 396%, from 40.067 to 198.611, indicating deeper and more frequent collaborative interactions, particularly in strategic emerging industries. These changes highlight the network’s transition toward greater efficiency, with enhanced local specialization and accelerated knowledge transfer across clusters (see Table 4).

(3) City centers exhibit volatility, with a core–periphery phenomenon. As seen in

Table 5. Top-10 cities by centrality in the intra-regional cooperative innovation network of the three northeastern provinces.

Indicator	Degree Centrality			Weighted Degree Centrality			Betweenness Centrality
Ranking	2009–2013	2014–2018	2019–2023	2009–2013	2014–2018	2019–2023	2009–2013	2014–2018	2019–2023
1	Shenyang	Shenyang	Harbin	Dalian	Shenyang	Shenyang	Harbin	Harbin	Harbin
2	Dalian	Changchun	Shenyang	Shenyang	Dalian	Dalian	Shenyang	Changchun	Jilin
3	Harbin	Harbin	Dalian	Qiqihar	Changchun	Changchun	Changchun	Shenyang	Changchun
4	Changchun	Dalian	Changchun	Anshan	Harbin	Harbin	Dalian	Dalian	Shenyang
5	Jilin	Jilin	Jilin	Changchun	Jilin	Jilin	Baishan	Jilin	Dalian
6	Fushun	Jinzhou	Anshan	Harbin	Qiqihar	Qiqihar	Jilin	Jinzhou	Anshan
7	Anshan	Qiqihar	Jinzhou	Jilin	Anshan	Anshan	Daqing	Jiamusi	Baicheng
8	Dandong	Jiamusi	Baicheng	Benxi	Jixi	Baicheng	Dandong	Qiqihar	Daqing
9	Liaoyang	Anshan	Chaoyang	Fushun	Yingkou	Fushun	Fushun	Qitaihe	Dandong
10	Daqing	Yingkou	Fushun	Dandong	Fushun	Huludao	Hegang	Daqing	Fuxin

, while the core cities, particularly those within the Shenyang–Dalian–Changchun–Harbin anchor cluster, continued to dominate the network, secondary hubs showed increased volatility. Harbin, for example, displaced Dalian as the leading city in degree centrality by 2019–2023, reflecting its growing importance in cross-provincial technology transfers, which led to a 214% increase in weighted degree centrality. Resource-based cities such as Daqing and Anshan retained their influence, while Baicheng gained prominence through agricultural innovation partnerships, increasing its number of collaborative ties by 17. Despite these shifts, a significant number of cities (12 out of 36) remained in the periphery, with low connectivity. This persistent inequality in connectivity suggests that the core–periphery structure of the network continues to be shaped by anchor-driven catch-up dynamics, with core cities serving as gatekeepers for the integration of peripheral cities, greatly impacting regional cooperation and the economic transformation of the northeast.

3.2.2. The Spatial Evolution Characteristics of the Inter-City Innovation Network Structures

(1) The cooperative innovation network among cities in the three northeastern provinces is primarily focused on intra-city cooperation, supplemented by inter-city collaboration. Over time, the intensity of both intra- and inter-city cooperation has increased (see Figure 4). Cities, such as Shenyang, Dalian, Changchun, and Harbin, possess strong industrial clusters and research institutions, while resource-based cities such as Daqing and Anshan have specialized industrial parks, forming tight-knit internal cooperation networks. For example, Shenyang leads in high-end manufacturing and innovative materials, while Anshan leverages local resources and special funds to create industrial parks for the development of new types of materials and energy batteries. However, inter-city cooperation remains relatively weak, limiting the overall regional innovation capacity and the pace of industrial upgrading. To promote the economic development and industrial transformation of the northeast, it is necessary to strengthen inter-city cooperation, establish cooperative policy frameworks, and provide financial support to facilitate the flow of technology, talent, and capital.

(2) The cooperative innovation network has gradually evolved into a polygonal spatial pattern. From 2009 to 2013, Shenyang and Dalian served as the core, forming a center–periphery structure, with the cooperation intensity between Dalian and Qiqihar being the highest. This was primarily due to Qiqihar First Heavy Machinery Group establishing a branch in Dalian, resulting in 124 cooperative patents and the maintenance of close innovation links. From 2014 to 2018, the network shifted to a multi-core structure, with strengthened connections among Shenyang, Dalian, Changchun, and Harbin, exhibiting a hub-and-spoke formation. Between 2019 and 2023, the network further developed into a highly complex polygonal structure. The core positions of Shenyang, Dalian, Changchun, and Harbin were consolidated, with their radiating influence on surrounding cities enhanced. The inclusion of peripheral regions made the network more comprehensive.

3.3. Analysis of the Mechanisms Influencing the Urban Cooperative Innovation Network

3.3.1. Model Test

On the basis of the above research into the characteristics of cooperative innovation networks, we next explore the evolution mechanism of cooperative innovation networks from the perspective of proximity.

The explanatory variable “strength of cooperative ties” was a count variable with a variance significantly greater than its mean, indicating overdispersion. In such cases, Poisson regression is inappropriate due to its assumption of equidispersion. The negative binomial regression model was selected as it accounts for overdispersion by introducing an additional parameter for variance. This model was preferred over other alternatives, such as zero-inflated models, as it effectively addresses overdispersion without unnecessary complexity. Additionally, prior to the regression analysis, we conducted a collinearity test, and the variance inflation factors (VIFs) for all variables were found to be less than 5, indicating no multicollinearity issues. Thus, the negative binomial regression model is a suitable and efficient choice for analyzing the data. The regression results are shown in

Table 6. Negative binomial gravity regression model results.

Variable	2009–2013	2014–2018	2019–2023
PATi	0.776 ***	0.891 ***	0.880 ***
	(0.071)	(0.059)	(0.058)
PATj	0.774 ***	0.889 ***	0.879 ***
	(0.072)	(0.059)	(0.058)
Geoij	−0.042	−0.151	0.354 **
	(0.261)	(0.196)	(0.174)
Borij	0.318	0.759 **	1.183 ***
	(0.348)	(0.302)	(0.240)
Ecoij	−1.358 **	−1.494 **	−1.630 ***
	(0.634)	(0.610)	(0.504)
Indij	1.420 **	2.260 **	−0.630
	(0.712)	(0.955)	(0.885)
Insij	1.027 ***	2.174 ***	0.925 ***
	(0.329)	(0.261)	(0.207)
Tecij	9.375 ***	−0.156	10.744 ***
	(3.012)	(1.472)	(1.848)
Socij·gov	0.135	−2.330 ***	0.561 *
	(0.578)	(0.485)	(0.333)
Socij·edu	−1.039	1.720 ***	1.323 **
	(0.805)	(0.626)	(0.546)
_cons	−18.748 ***	−13.530 ***	−22.027 ***
	(2.614)	(1.611)	(1.965)
N	812	1056	1122
Alpha	3.047	2.975	3.657
	(0.402)	(0.317)	(0.081)
Prob > chi2	0.0000	0.0000	0.0000
Log likelihood	−599.29705	−865.75015	−1471.6152
Pseudo R2	0.2232	0.2307	0.1685

Note: * p < 0.1, ** p < 0.05, *** p < 0.01, values in parentheses are standard errors.

.

3.3.2. Empirical Research Results

(1) The innovation capacity of cities, represented by PAT_i and PAT_j, remained significant at p < 0.01, with a positive regression coefficient. This aligns with the recombinant innovation theory, where cities with higher patent stocks exhibit stronger abilities to absorb and recombine cross-regional knowledge [36]. The three northeastern provinces’ “14th Five-Year Plan for Digital Economy” (released in 2021) has accelerated this process through cloud-based innovation platforms. The three northeastern provinces of China have established a strategic emerging industry collaboration network centered on technological innovation through policy-driven initiatives and industrial transformation, significantly contributing to the region’s sustainable development.

(2) The impact of geographical proximity on collaborative innovation networks has undergone significant changes over time. The spatial distance had a significantly positive regression coefficient between 2019 and 2023, and boundary proximity was significantly positive in both 2014–2018 and 2019–2023. This indicates that in the early stages, innovation activities primarily relied on the flow of technology and resources rather than geographical proximity. As the industry matured, the importance of geographical proximity increased, reducing collaboration costs and promoting inter-city cooperation and technological innovation [37]. This finding refutes the “death of geographical proximity” hypothesis [38].

(3) In terms of economic proximity, the impact coefficient of internal innovation linkages was significantly negative across all three periods. This indicates that substantial economic disparities between cities can foster innovation cooperation [39]. Notably, complementary cooperative relationships have emerged among economic-center cities and resource-based cities. Resource-based cities provide raw materials and resources, while economic centers offer technological and knowledge support. Together, they drive balanced regional economic development and industrial upgrading.

(4) In terms of industrial proximity, this significantly promoted inter-city cooperation between 2009 and 2013 and between 2014 and 2018 but had a negative and insignificant effect between 2019 and 2023. This indicates that in the early stages, cities with similar industrial structures found it easier to collaborate. However, in the later stages, local protectionism and competition factors limited the flow of innovation elements, thereby inhibiting regional innovation linkages. Consequently, in recent periods, cities have sought cross-industry cooperation to achieve technological breakthroughs [40].

(5) In terms of institutional proximity, this has consistently and significantly positively impacted inter-city cooperation. Similar policy environments and institutional frameworks reduce cooperation risks and barriers, facilitating rapid collaboration. This strengthens regional economic integration and technological synergy, driving economic development and technological innovation [41].

(6) In terms of technological proximity, this significantly promoted inter-city cooperation between 2009 and 2013 and between 2019 and 2023, while having an insignificant negative impact between 2014 and 2018. In the early stages, the role of technological similarity in reducing communication costs and facilitating knowledge exchange is consistent with Boschma’s proximity framework, which highlights technological proximity as a critical driver of collaborative innovation [17]. In the mid-term, the shift toward technological diversification aligns with Frenken et al., who argue that regions often engage in “related variety” to avoid lock-in effects and explore new innovation pathways [42]. Balland et al. further emphasize that excessive technological proximity can stifle novelty, necessitating strategic differentiation [43]. In the later stages, policy guidance enhanced technological sharing and knowledge transfer between cities, strengthening cooperation and driving technological innovation and economic transformation [44].

(7) In terms of social proximity, proximity in scientific expenditure and educational expenditure had different impacts on collaborative innovation networks. Initially, scientific expenditure proximity had a significantly negative effect, indicating that large differences in government support between cities promoted the flow of innovation elements from economic center cities to resource-based cities. This “compensatory diffusion” is consistent with Breschi and Lissoni, who argue that knowledge spillovers often follow gradients of resource inequality [45]. In the later stages, this turned positive, suggesting that inter-city cooperation relied increasingly on government funding support [46]. Educational expenditure proximity initially had an insignificant impact on the cooperation network, but as educational disparities decreased over time, it promoted inter-city innovation cooperation, strengthened knowledge flow and technology exchange, and provided a solid foundation of human and knowledge resources. Glaeser et al. argue that homogenized educational investments reduce cognitive barriers, enabling cities to share tacit knowledge and co-develop technologies [47].

4. Discussion

4.1. Analysis of Influencing Mechanisms in Urban Cooperative Innovation Networks in the Three Provinces of Northeast China

The aforementioned analysis explains the evolutionary mechanisms of the innovation network in the three northeastern provinces from the perspectives of urban innovation capacity and multidimensional proximity. Urban innovation capacity, economic proximity, and institutional proximity have consistently and significantly influenced the evolution of the innovation network. In contrast, geographical, industrial, technological, and social proximity have shown variations in their effects on network evolution. Together, these factors have jointly promoted the evolution of the innovation network in the northeastern provinces (see Figure 5).

Additionally, the external innovation network of the three northeastern provinces had an important impact on the internal network. The external network, characterized by strong innovation ties but low density, has enhanced the technological level and innovation capacity of enterprises and research institutions within the internal network through the knowledge spillover effect. The interaction between the external and internal networks has not only facilitated the sharing of technology and resources but also strengthened the competitiveness of the internal network. Over time, the internal network has continuously absorbed the advanced achievements of the external network, thereby improving its own innovation capacity and market competitiveness.

4.2. Recommendations

4.2.1. Institutionalize Cross-Regional Collaboration Mechanisms

A Northeast Regional Innovation Coordination Platform should be established to integrate digital resources across municipalities within and beyond the three northeastern provinces (Liaoning, Jilin, and Heilongjiang). This platform would facilitate efficient resource sharing and information exchange among various entities. Additionally, a dedicated Interprovincial Innovation Fund should be created to support collaborative R&D projects between the northeastern provinces and major economic hubs, such as Beijing, Shanghai, and Tianjin. To further strengthen regional ties, an annual Northeast Strategic Industries Summit should be initiated, focusing on fostering partnerships in key sectors, such as advanced manufacturing, clean energy, and smart agriculture.

4.2.2. Optimize Spatial Development Patterns

A formal quad-city coordination framework should be established among Harbin, Changchun, Shenyang, and Dalian. This framework would include joint spatial planning committees for industrial zoning, cross-jurisdictional infrastructure investment proto-cols, and specialized transition funds to support resource-depleted cities like Fuxin and Daqing. To enhance connectivity, a high-frequency inter-city rail network should be expanded, and unified logistics data platforms should be developed to integrate supply chains. Furthermore, innovation voucher programs could be implemented to link R&D institutions in core and peripheral cities, promoting knowledge transfer and collaborative innovation.

4.2.3. Strengthen Regional Innovation Network Through Multidimensional Proximity

To advance urban innovation strategies and sustainable development, prioritize multidimensional proximity (institutional, technological, industrial) in Northeast China’s regional planning. Anchor growth in the Harbin–Changchun–Shenyang–Dalian metropolitan cluster, fostering cross-city innovation corridors and green industrial hubs. Optimize sustainable industrial clusters (e.g., clean energy, smart manufacturing) with shared infrastructure to reduce ecological footprints. Establish a unified factor market to accelerate flows of talent, capital, and green technologies, supported by aligned regulatory frameworks and carbon-neutral incentives. Strengthen institutional proximity through joint R&D platforms and circular economy policies. Introduce flexible talent-sharing mechanisms, prioritizing skills in low-carbon sectors, while expanding cross-regional public transit to enhance connectivity.

4.3. Limitations and Prospects

4.3.1. Limitations

While the analysis of jointly filed patents (n = 26,913) effectively reveals formal innovation collaborations between cities, this methodological focus necessitated the exclusion of 389,355 individual patents (representing 6.91% of total patents) from the original dataset. These excluded patents, while not indicative of inter-organizational collaboration, may contain valuable insights into intra-organizational innovation patterns and solo inventor activities. Future investigations could employ alternative methodologies, such as citation network analysis or inventor mobility tracking, to extract latent collaboration patterns from this substantial body of excluded data.

4.3.2. Prospects

(1) Multi-Actor Collaboration Dynamics: Subsequent studies should systematically examine the triadic relationships between enterprise–university–research institute consortia, particularly analyzing how funding mechanisms influence collaboration patterns in advanced manufacturing and new energy sectors. Comparative analysis of collaboration efficiency metrics across different institutional combinations would yield actionable policy insights.

(2) Industry-Specific Network Topologies: The nine strategic emerging industries (including new-generation information technology and high-end equipment manufacturing) require discrete network analyses using multiplex network modeling. This approach would enable the identification of industry-specific knowledge flow channels, cross-sectoral technology convergence points, and critical path dependencies within regional innovation value chains.

5. Conclusions

This study is based on the PatSnap patent database to construct and map strategic emerging industry innovation networks among cities within the three northeastern provinces of China from 2009 to 2023. It employs GIS 10.8 spatial visualization and Gephi10.1-based topological analysis to characterize spatio-temporal evolution patterns and network structural dynamics, while applying a negative binomial gravity regression model to quantify influencing factors and elucidate the spatial interaction mechanisms of innovation networks. The main conclusions are as follows:

(1) The external innovation network of the three northeastern provinces has been expanding in scale, but its density has been decreasing year by year. The newly joined cities do not have close cooperation with the northeastern provinces, leading to a dilution of the network. We believe that this characteristic of the external collaborative innovation network does not meet the current economic development needs of the northeastern provinces. Simply expanding the network’s scale without increasing its density is unlikely to effectively enhance the economic competitiveness of the region.

(2) The external innovation network of the three northeastern provinces is evolving into a pattern of “dense in the southeast and sparse in the northwest”, resulting in an uneven spatial distribution. Cooperative relationships are concentrated on the eastern side of the “Hu Huanyong Line”, with Beijing serving as the hub and municipalities directly under the central government, provincial capitals, and sub-provincial cities acting as the main cooperation nodes.

(3) The internal innovation network of the three northeastern provinces has become increasingly compact over the years, exhibiting small-world characteristics and forming a “core-periphery” structure. The four major cities of Harbin, Changchun, Shenyang, and Dalian remain at the core. The secondary core resource-based cities have experienced fluctuations; however, they continue to be crucial nodes for industrial development and coordinated transformation. The innovation network has evolved from a simple core–periphery structure to a multi-core hub-and-spoke network, eventually forming an extensive polygonal network structure. From an internal perspective, the development pattern of the cooperative innovation network in the northeastern region aligns with general trends. However, it is noteworthy that the intensity of cooperative innovation between cities is still relatively weak compared with the more developed regions in Southern China. While focusing on optimizing the internal structure, it is essential to enhance the intensity of cooperative innovation to drive the sustained and rapid development of strategic emerging industries.

(4) In the impact mechanism on cooperative innovation networks, urban innovation capacity has consistently played a strong and enduring role in the evolution of the innovation network, while the effects of multidimensional proximity indicators have fluctuated over time. Institutional proximity has a significantly positive impact, as cities with similar institutional environments are more conducive to collaborative innovation. Economic proximity generally has a positive effect, facilitating inter-city cooperation. However, the close cooperation between economic-center cities and resource-based cities in the northeastern provinces exhibited a negative effect; greater economic disparities stimulated stronger cooperative relationships between cities. Geographical proximity only showed a positive impact in the later stages, indicating that it gradually became an important factor in promoting inter-city collaborative innovation. The impacts of industrial, technological, and social proximity varied, but overall, they exhibited a “proximity connection” network growth trend. The external innovation network supported the development of the internal network through knowledge spillovers and resource sharing. This was particularly evident in the northeastern provinces, where it promoted the evolution of the urban collaborative innovation network.

This study significantly deepens our understanding of the major city-level collaborative innovation networks within the strategic emerging industries in the three northeastern provinces of China. Through a comprehensive analysis of the spatio-temporal evolution patterns, network structural dynamics, and influencing factors of both external and internal innovation networks, it offers profound insights into how major cities can collaborate more effectively in the innovation process. The collaborative innovation networks among major cities play a crucial role in driving the sustainable development of these cities. This kind of innovation-led cooperation can help cities reduce resource waste and optimize industrial layouts, both of which are essential for long-term sustainable development. Building on these findings, future research could further refine our understanding of the innovation network by considering the following: (1) Individual patent applicants can broaden the scope of their research investigations by employing other methods, such as citation and inventor mobility tracking, to extract potential patterns of collaboration from previously excluded large-scale datasets. (2) Tripartite interactions within industry–university–research consortia should be systematically examined to explore their collaborative dynamics. (3) Nine strategic emerging industries (including next-generation information technology and high-end equipment manufacturing) require discrete network analysis through multi-network modeling approaches to identify domain-specific knowledge transfer pathways.