Innovation ecosystems (IE) have gained significant attention in the past decade [1] due to the pivotal role of innovation in implementing sustainability [2,3,4]. Recognized as a new driver of economic growth and a pathway to sustainable competitive advantage [5,6], scholars have sought to define IE concepts and components through literature reviews and qualitative research [7,8]. However, given diverse research focuses and theoretical foundations, the study of IE frameworks is currently marked by “diversity and contention”. The composition and boundaries of IE have yet to be unified and established through quantitative research [9].

The subject of innovation, such as scientists, inventors, companies, research institutions, government departments, and investors, drives and influences the development of innovation through various interactions. Interactions, which encompass collaboration, competition, coopetition, and more, serve as the core driving force for promoting innovation. And the object of innovation, including knowledge, products, and services, represents both the resources and outcomes of innovation. Hence, at the conceptual level, based on quadruple and quintuple helix innovation systems, the innovation ecosystem is complex and heterogeneous [10], comprising actors, activities, and artifacts [7].

Network science serves as a crucial perspective for studying the innovation ecosystem, acting as a driving force for innovation development [11,12]. However, existing research on innovation networks commonly utilizes single-layer or homogeneous networks [13,14], which presents certain limitations in characterizing and describing the complexity and heterogeneity of the innovation ecosystem framework [15,16]. It is necessary to introduce multi-layered and heterogeneous networks for future inquiries into its structure and patterns.

Moreover, the utilization of multi-source heterogeneous data facilitates accurate identification and in-depth exploration of the distribution and proximity of actors and artifacts, interconnected through various activities. This approach establishes a robust data foundation and technical support for uncovering the collaborative mechanisms within the innovation ecosystem [17].

The innovation ecosystem concept arises from the integration of “business ecosystems” [18] and value co-creation process [19]. It has become a new research perspective in innovation and innovation management [8,20,21].

Etzkowitz and Leydesdorff (2000) proposed the Triple Helix innovation model consisting of university–industry–government, emphasizing the importance of knowledge generation and innovation in the economy [22,23]. Subsequently, Carayannis and Campbell (2006) added the ‘media-based and culture-based public’ and ‘civil society’ as a fourth helix to the Triple Helix model, forming the Quadruple Helix innovation model [24]. This model highlights the need for the sustainable innovation of knowledge in the economy to evolve in conjunction with knowledge society and democracy. Furthermore, considering the importance of the ‘natural environments of society’, Carayannis, E.G., et al. (2012) proposed the more comprehensive Quintuple Helix innovation model [25,26]. This model emphasizes the crucial role of socio-ecological transformation in driving innovation. Therefore, innovation ecosystems, based on both democracy and ecology, exhibit broader scope, increased complexity, diversity, and heterogeneity.

Researchers defined the composition and boundaries of IE architecture according to different research preferences and goals [9] through literature review and bibliometric studies [27,28]. Based on previous research, Granstrand and Holgersson (2020) [7] proposed the “3A” architecture of innovation ecosystems, including participants, artifacts, and activities. However, quantitative research in this area is limited, with most studies being theoretical. 

Innovation actors collaborate to form social networks, while knowledge elements combine to form knowledge networks. Both are innovative interactive networks. Previous social network research has explored cooperative relationships, information flows, and decision-making processes among countries, organizations, and individuals, respectively [29,30,31]. However, a comprehensive analysis of innovative social networks at these levels is lacking. Knowledge elements are the preliminary conclusions held by research communities in scientific and technological fields [32]. They encompass facts, theories, methods, and procedures related to specific topics [33,34]. The innovation of emerging technologies relies on combining or recombining these knowledge elements, making them crucial for innovation [34,35]. In fact, existing knowledge network research has divided knowledge elements into scientific knowledge and technical knowledge. Scientific knowledge elements focus on knowledge discovery [36], while technological knowledge elements emphasize ownership and application [37]. Both types of knowledge are important in the innovation ecosystem [32,38] and are interconnected with the relationships between innovation actors. Several studies have examined networks of scientific knowledge elements [39,40], while others have focused on networks of technological knowledge elements [27,41,42]. Previous studies have examined scientific knowledge element networks [39] or technical knowledge element networks [17,43].

Research specifically analyzing and comparing these two types of knowledge meta-networks is still limited. The network characteristics of scientific and technological knowledge need to be integrated and mined. Furthermore, collaborative interactions among innovators [44] and their impact on the potential to combine scientific and technological knowledge are key factors in promoting innovation. 

The social network formed by innovators and the knowledge network composed of knowledge elements have a double embedded relationship [32]. At the same time, they exhibit heterogeneity and decoupling [15]. Previous research has explored the impact of dual embedded relationships in heterogeneous networks on organizational innovation [45], emphasizing that the degree of integration of social networks and the diversity of knowledge networks are two driving factors for organizational-level innovation. Over time, they evolve in a spiral pattern [26]. Thus, the IE construction includes two heterogeneous networks: social network and knowledge network.

Innovation collaborations can be organized and managed at multiple levels of analysis [29] with a multilevel network perspective [46,47]. However, most studies on innovative complex networks focus on single-layer and homogeneous networks, ignoring the underlying characteristics of real complex innovation ecosystems. Combining the above studies and the diversity and heterogeneity of innovation ecosystems, it becomes imperative to utilize multi-layer heterogeneous networks [16,48] to study and understand the IE framework.

Furthermore, utilizing multivariate heterogeneous datasets allows for improved resolution and consideration of multiple characteristics of real innovation ecosystems [49]. Bibliometrics [50] and patent data [27] have been widely used in innovation research to assess and characterize technological innovations. Researchers integrated different heterogeneous data to improve the validity and credibility of the study [16] and analyzes a novel IE framework for emerging technologies [51].

An innovation ecosystem niche, as a fundamental element of the structuralist approach to ecosystems, refers to the specific position of actors and artifacts within the flow of overall system activity [5]. It is related to strategic resources and direction [52] and has a positive impact on the exploration and utilization of innovation [53]. Assessing the positioning of innovation subjects and products in the ecosystem and distinguishing the focus subjects and internal logic of the innovation ecosystem are of great significance to value creation [36].

Hub nodes and bridging nodes occupy special ecological niches in the ecosystem framework [53]. Hub nodes have a large number of connections and play a critical role in strong connections. Collaboration between Hub nodes drives technological progress, promotes cooperation and standardization, identifies key areas and innovators of the technology ecosystem, and guides and supports technological development and innovation. On the other hand, bridging nodes have the ability to bridge different sub-networks in weak connections, facilitating information transfer and facilitating collaboration between different actors or knowledge areas. Network features, degree [32], and betweenness centrality [54] are used to identify these nodes.

Knowledge proximity, which focuses on the similarity of knowledge structure and technological experience among innovation actors, has received significant attention in recent innovation network research [55,56]. Exploring the proximity of knowledge in the innovation ecosystem is beneficial to the interactive learning, knowledge acquisition and innovation success of innovative subjects.