The future high-performance social network is based on a group of high-quality node users who actively participate in the social platform first. The continuous and active participation of high-quality node users makes their influence multiply exponentially. Among them, there are a large number of high-quality node users in industry-university-research social activities, which become the "metaverse-like" social circle that high-performance social networks can build in the future based on the current 5G network communication infrastructure.

Industry-university-research social activities can not only effectively promote the development of cutting-edge science and global technological innovation, but also help shape the careers of scientific individuals. The establishment of open science infrastructure such as industry-university-research social websites, open access platforms, and institutional knowledge bases has lowered the threshold for scholars and other members of the public in various disciplines to participate in the scientific agenda [1], and has also become a new "public domain" for knowledge diffusion and meaning production.

Among them, industry-university-research social websites, with their advantages in providing a large number of free academic resources, can help scholars promote scientific research results and accumulate social capital in vertical fields, and therefore have won the recognition and participation of many researchers around the world. It is well known that user interaction relationships, information content, and user influence [2] are the three core sources of value of the new generation of the Internet. The individual relationship networks built by a large number of users on industry-university-research social websites will play an important role in scientific communication. Scholars generate corresponding social relationship networks by capturing the social data of users on industry-university-research social networking sites, analyzing the network topology and network characteristics of different users [3-4], or based on this, analyzing the social contributions or values of different members, the types of relationships between members [5-6], communication patterns and intensity [7], and the dissemination paths of information. However, there is a lack of explanatory research on the causes and effects behind the network structure. Therefore, this paper adopts a mixed research method, conducts in-depth interviews based on objective data analysis, and asks the builders to make subjective interpretations of their social relationships in order to deeply understand the online scientific communication behavior of scholars. As the three giants of social networking sites, ResearchGate, Academia.edu and Mendeley have most of the world's scientific research users, especially high-quality scientific researchers. On the one hand, the academic discourse system is still mainly in the hands of developed countries, and English is the main communication method; on the other hand, my country's vertical industry-university-research social market focusing on a certain field is still relatively blank. Platforms such as Keyi.com, Xiaomuchong, and ScienceNet cannot gather high-quality users at all due to poor user experience and lack of academic resources. According to existing data statistics, as of February 2022, more than 20 million users have shared, discovered and discussed academic research on ResearchGate, and the proportion of registered Chinese researchers ranks second only to the United States; the number of shared documents on Academia.edu exceeds 28 million, covering major natural sciences, and is an important platform for the circulation of cutting-edge scientific research information; Mendeley has hundreds of millions of academic achievements. Taking into account the platform's functions, number of users, and data availability and integrity, this article selects ResearchGate, the most commonly used industry-university-research social networking site, as the source of individual network data.

1. Structural Holes and Intermediary Role Theory

In 1992, Burt proposed the "Structural Holes" theory in his book Structural Holes: The Social Structure of Competition [9]. Since then, this theory has been widely used in various fields of social sciences. Its application in social network research is mainly concentrated in the following two aspects: first, exploring the embedding and effectiveness of structural holes in social networks; second, identifying key subjects of information dissemination (key subjects of social network information dissemination, core authors in subject areas, and high-impact journals) [10]. Overall, existing research has not yet revealed the social network structure of industry-university-research social users in depth. What impact does the position of network members have on the acquisition and dissemination of knowledge? What are the common characteristics of key network figures? And how do they influence other members in the social network? These issues need to be further explored in practice.

Based on this, this paper draws on the theory of structural holes and intermediary roles, constructs a measurement and analysis framework, and uses researchers and related data on three major industry-university-research social networking sites as analysis samples. It attempts to identify the positions, role types, and roles of different personnel in the industry-university-research social network, and then explores how key users can influence the circulation of academic information, in order to provide useful references for related theoretical research and practical applications.