CC BY
58
Georgy Nikolaenko
MA in Sociology, Research Fellow
S.I. Vavilov Institute for the History of Science and Technology of the Russian Academy of Sciences, St Petersburg Branch,
St Petersburg, Russia; e-mail: eastrise.spb@gmail.com
Roman Malyushkin
MA in Computer Science, Independent Researcher
St Petersburg, Russia; e-mail: malyushkinr@gmail.com
Anna Samokish
PhD, Research Fellow, S.I. Vavilov Institute for the History of Science and Technology, of the Russian Academy of Sciences, St Petersburg Branch,
St Petersburg, Russia; e-mail: tomasina84@mail.ru
Global Distribution of Digital Scientific Communication: Сase of ASNS ResearchGate
УДК: 316.77
DOI: 10.24411/2079-0910-2020-13011
The paper describes the first stage of a comprehensive study of the academic social network ResearchGate.net. The objectives are to formulate a theoretical basis for the research of Academic Social Networking Sites (ASNS) and to analyze the first data of the global geographic and disciplinary distribution of ASNS ResearchGate.
The theoretical concept involves the analysis of academic social networks as complex socio-technical systems for the production, distribution and accumulation of scientific information. Hence, academic social networks are considered both as an environment and as an instrument of scientific communication. As an environment, ASNS are defined as a quasi-virtual field of scientific interaction, where "active" and "passive" types of communicators are distinguished. In turn ASNS as an instrument is a comprehensive system for archiving scientific information, indexing content, and
© Николаенко Г.А., Малюшкин Р.В., Самокиш А.В., 2020
ensuring information flows. Particular attention is paid to the ratio of formal and informal scientific communication in the context of information glut, which determines new types of practices. The focus is on the macro-characteristics of the platform — the distribution of all users by country and scientific discipline. The data obtained are visualized with the mapping method and the form of multivariate graphs made by the method of proportional areas. The paper reports the results of pilot research, which allows to study the social dynamics of ResearchGate academic social network. Particular attention is paid to testing the trends of disciplinary homogenization and social globalization evidence from this site.
The attained results provide the broadest picture of the ResearchGate social structure at the moment. The characteristic of the general population, in contrast to the use of local samples, allows to overcome a number of limitations associated with representativeness. The developed research design made possible to fix the semi-annual global dynamics of the social structure of the network, which allows us to test local hypotheses throughout the platform. We also believe that such research design will provide a number of new questions and results, which will increase the heuristic potential of studying scientific communication on the Internet.
Keywords: ASNS, ResearchGate, network, academic communication, social structure, statistics, webmetrics, altmetrics, longitudinal studies, academic social network sites, academic networks, globalization.
Introduction
More than 10 years have passed since the appearance of the largest academic social networks (ResearchGate.net and Academia.edu). Most of the scientists see them as an integral part of any scientific work. We can admit, that Academic Social Networking Sites (ASNS) have become the new players in the scientific communications market. The specifics of science, in particular the network nature of the intrascientific communication, ensured a rapid transition of new communication tools, significantly expanding the geographical coverage of communication and its intensity.
We emphasize that we are talking specifically about social networks targeted at scholars and academic information exchange, and not about networks aimed at the general public (LinkedIn or Facebook) in which, undoubtedly, many researchers are also registered and they also have scientific communication within.
We consider ASNS as "web-based services that allow individuals to (1) construct a public or semi-public profile within a bounded system, (2) articulate a list of other users with whom they share a connection, and (3) view and traverse their list of connections and those made by others within the system" [Boyd, Ellison, 2007, p. 211].
Researchers use the capabilities of ASNS as reputational mechanisms, providing their visibility in the conditions of information overload. In addition, ASNS are used to varying degrees by different institutions both to develop a strategy for their scientific activities and to evaluate the effectiveness of employees. Thus, ASNS internal indicators become a kind of altmetrics that can complement the currently used bibliometric indicators and hence increase the reliability of evaluating the effectiveness of scholars and scientific institutions.
The leaders among ASNS at the moment are ResearchGate.net and Academia.edu, uniting millions of users from around the world. Both are for-profit venture capital-funded technology startup companies [Jordan, 2019, p. 2]. However, the monetization models of these networks are different. Many of the services provided by the Academia.edu (like "See the papers that mention you" or "Measure the full impact of your research") are paid and
thus put users in a precarious position. We see the study of the structure of the ResearchGate (although this network is inferior to the first in the number of users) to be more promising. On the first hand it provides a wide range of tools for users, all of them are free and thus equalizing users' opportunities. In addition, registration filters are stricter, confirmation of academic affiliation is required. Consequently, a sample of ResearchGate (RG) users will be more indicative than Academia.edu precisely from the point of view of academic information exchange and communication, since it includes representatives of science, education and science-based business only.
We see it useful to study the macro-characteristics of the network, namely the geographical and disciplinary distribution of users, in order to understand whether these factors affect user practices, the dissemination and exchange of scientific information, and other processes. In order to see the detailed characteristics and trends within the network and individual groups, we need to analyze the initial macro-characteristics and their dynamics.
Related Research
The growing popularity of ASNS has obviously aroused interest among researchers in issues related to these sites and the emergence of large number of studies. However, most of them are local, made on limited samples (within the same institution, city or country) and, as a rule, they do not offer any global hypothesis.
The most complete picture of research on ASNS was presented by Katy Jordan in her review of the history and scholarship of academic social network sites in 2019 [Jordan, 2019]. The author focuses on the empirical studies on various aspects of ASNS in order to address the role that they play, their benefits and limitations. She noted that more than 10 years have passed since the creation of the ASNS, and it requires the careful analysis of research that has been carried out during this period and a revision of the ASNS concept as "Facebook for scientists". K. Jordan is one of the few researchers who proposed a theoretical concept for the development of academic networks, as a transition from a network for communication and collaboration to a platform for publishing research materials. She singled out the absolute leaders — the reception gate and the Academy, and primarily focused on the studies dedicated to them.
Researchers are primarily interested in the academics' practices in using ASNS: the reasons for choosing one of the sites, the purposes and habits of using the network functions, fears and concerns during work, the frequency and duration of working in ASNS, etc. Another important question posed by researchers is the possibility of using ASNS as a source of altmetric indicators. Altmetrics in this context are considered as a replacement or rather an addition to traditional bibliometric indicators, the source of which are various Internet resources, including ASNS. The correlation of ASNS metrics with traditional bibliometric indicators, the publication activity of users in ASNS, justified by the strategies of researchers in an effort to increase indicators and therefore their prestige, is one of the most discussed topics. Jordan wrote that "metrics is the most prevalent theme within the body of literature related to ASNS" [Jordan, 2019, p. 6].
A range of studies have shown the benefits of ASNS for researchers and their institutions to build their international reputation.
In this paper we would like to single out some papers devoted directly to the geographical and disciplinary structure of ASNS. We should note that we can rely on certain quantitative
results only in terms of tracking the dynamics of the social structure, but this is quite difficult, since most studies are not longitudinal and do not overlap one another. Nevertheless, they allow us to elicit some trends that we intend to test with the empirical research.
Despite the time elapsed since the publication, one of the most relevant is «Online collaboration: scientists and social networks» [Van Noorden, 2014] based on the questionnaire in «Nature». Because of the fact that it was based on a sampling frame of "Nature" readers, we can't be sure in its representativeness in the field of disciplinary differences. Nevertheless, the study has shown that researchers from different scientific fields use social networks differently. According to Van Noorden's statistics ResearchGate and Academia.edu were mainly used for academic collaboration and information exchange. The trends identified by the author were confirmed by further studies on local samples in various institutions and countries.
An analysis of local studies (putting them together and comparing) might allow to try building a map of the RG use, but we cannot admit even 10% of countries or institutions are covered with such studies. The local studies which are interesting for us can be divided into several groups: those dedicated to the choice of a particular network among representatives of a particular country and those dedicated to the analysis of the disciplinary affiliation of users of a particular network (we were more interested in the ReserchGate network) from any institution or country.
Among the first, we can mention an internal study conducted at the University of Bergen (Norway) [Mikki et al., 2015]. Among the questions the researchers were asked we could see the following: "What services are the most popular and why?" or "Does gender, age, position or faculty play a role?" It was shown that 76% of researchers were registered only at ResearchGate.net, so this platform can be called the most important for the University of Bergen. Nevertheless, we cannot extend these conclusions even to Norwegian scientists in general, although the University of Bergen is the second largest university in the country and therefore one of the most involved in the international collaboration. Thus, we cannot say anything about scientists from smaller institutions.
A study carried in Japan [Mason, 2020] demonstrated that Japanese user activity on Academia.edu is very low, and only 30% of the sample are registered at ResearchGate. So the author indicates moderate use of the platform by Japanese academics. Altmetric analysis has shown that the use of ResearchGate was largely passive, and interactive features that could facilitate interaction with international researchers were not used.
As for the second group, one of the most important studies dedicated to the distribution of scientists on social networks by scientific discipline was carried out in several stages by J.L. Ortega [Ortega, 2015; 2017] in CSIC (Consejo Superior de Investigaciones Científicas, Higher Council for Scientific Research of Spain). The first findings on the ResearchGate network he made back in 2015 in the article "Disciplinary differences in the use of academic social networking sites" [Ortega, 2015], where he presents the results of his research on a sample of researchers working in CSIC institutions, aimed at identifying the possible impact of the discipline of researchers their preferences regarding Internet sites for scientific communication. The study was conducted in 2014-2015: using SQL scripts, indicators of 7193 profiles belonging to 6206 researchers from CSIC were obtained. He analyzed six quarterly samples from April 2014 to September 2015, characterizing the user activity of CSIC employees in ASNS-environments to identify possible changes in the disciplinary structure within the three main academic social networks — ResearchGate, Academia.edu and Google Scholar Citations. It immediately revealed two social phenomena related to
the influence of researchers' disciplinary affiliation on their communicative practices — in particular, the disciplinary imbalance present at the time of the study within various ASNS (as noted by Van Noorden and some other authors who demonstrated the predominance of representatives of various disciplines depending on the site). For example, the study conducted at the University of La Coruna [Fernandez-Marcial, Gonzalez-Solar, 2015] obviously showed the predominance of representatives of the natural sciences among RG users from this university.
Ortega showed that the majority of Academia.edu users came from the humanities, while the bulk of the RG was made up of representatives of biomedical sciences. It was also found that users from the humanities, social sciences, and natural sciences interact more actively and exchange information with others, while biology and biomedicine researchers tend to use RG more passively. However, at the same time, Ortega noted the process of gradual disciplinary homogenization of ASNS users [Ortega, 2017]. He showed that despite the statistical superiority of STEM disciplines in ResearchGate and the humanities in Academia. edu, the growth observed in both sites is due to the opposite directions (for Academia. edu — due to chemists, biologists and physicians, while for RG — due to Humanities and Social Sciences). So we can suppose that over time the differences in disciplinary structure observed at different networks may be balanced, as growth of the initially well represented subjects slowed down in the sample while growth increased in underrepresented areas. This trend is also confirmed by other studies [Alarcon-del-Amo et al, 2015].
In 2016 Ortega published a book dedicated to social networks for scientists [Ortega, 2016]. It also presents statistics on the disciplinary distribution of publications both within networks (in particular, ResearchGate and the Academia.edu), as well as a comparison of the main ASNS by dominant disciplines. The author identifies the Academia and the RG as a document sharing systems, which seems to us to be a rather strange restriction for these networks. This is probably why he analyzes the disciplinary distribution specifically for publications, not user profiles (in contrast to the mentioned articles).
A study by M.Thelwall and K.Kousha from the University of Wolverhampton has shown the up-and-down statistics of the users' activity from different countries and associates those metrics with the publication rates in WoS [Thelwall, Kousha, 2017]. The paper was published in 2013, when the number of users was RG was much lower than now. The statistics on publications and on the network of ResearchGate users were analyzed (a similar study was made on the Academia.edu [Thelwall, Kousha, 2014] and it was revealed that the published leaders were the USA, Japan, Sweden and Canada.
The next publication on ASNS which is interesting for us is not about researchers anymore, but about publications and their distribution on the ReserchGate. Thelwall and Kousha showed that the largest share among the articles uploaded to the network is related with biomedical topics, which is most likely due to the predominance of researchers in this discipline [Thelwall, Kousha, 2017]. They are considering a large-scale ad hoc sample of 68,731 publications in ResearchGate. The analysis shows that there are disciplinary differences in the degree of article sharing through a platform with wider coverage of the natural and physical sciences in comparison with the social and humanities. There is also a temporary effect in which there is a much wider distribution of publications in recent years.
Having studied empirical studies related to the geographical and disciplinary distribution in the ASNS, we can assume that a large number of gaps remain in the field of global research, as well as attempts to build a single conceptual model for the development of these platforms. J. Komljenovic came to similar conclusions: "what is missing are
methodologies that bring together the quantitative analyses that are possible with large data sets extracted from the academic platforms and the theoretical interpretive frameworks of education studies and related disciplines like sociology of education or political economy of education. In other words, there is room for empirical studies that make use of large digital data sets to theorise the insights with specific interpretive frameworks" [Komljenovic, 2019].
It confirmed our intentions to conduct an empirical study of the global structure of the ASNS of ResearchGate, with an attempt to theoretically interpret the network mechanism itself and test hypotheses of disciplinary and geographic homogenization.
Theoretical and methodological aspects of studying academic social media
We believe that the development of a conceptual sociological scheme of academic social networks will significantly increase the heuristic potential of any empirical studies related with these platforms. However, such development is not possible outside the context of scientific communication and information exchange research. We see the Grounded Theory as the most effective form of scientific understanding of ASNS, which implies the alternation of the collection of research material and theoretical conceptualization. Thus, our study includes both elements, and our conceptual framework, based on classical studies of scientific communication, is, in our view, somewhat ahead of empirical analysis, thereby fulfilling the role of a "spotlight". Of course, this imbalance is not large enough to affect negatively causal research schemes, and, the conceptual scheme can and will be corrected in accordance with the data obtained.
Academic social networks can be considered in the context of coping strategies for the information crisis of scientific communication, called in the literature "information explosion" [Barnett, 1964; Green, 1964] In particular, we propose to regard the academic social networks in a somewhat twofold manner, as an environment and as a means of scientific communication, which allows to overcome the limitations of classical systems of scientific communication.
As an environment, ASNS appears as a quasi-virtual field of scientific interaction, which includes two main types of components — "active"(AC) — users and scientific materials (including not only publications in the classical sense of the word, but also presentations, preprints , data arrays, and so on) and "passive" (PC), namely, user-generated content and self-generated content. The former is capable of active communication through the network, while the latter acts as a backdrop accumulating digital traces of "active" scientific communication. So, for example, the background includes comments, questions and answers in the "Q&A" section, lists of recommendations and various metrics, which can be bibliometric indicators or web metrics.
A direct comparison of the "active" and "passive" communication models seems to us to be a dead end. It is the expanded understanding of the term "communicator" in ASNS that is really important. A text published and indexed in academic networks is endowed with the ability of independent "existance", limited only by the algorithms of the environment. In other words, the text builds its own network of interaction with:
• Researchers (AC) through readings, downloads and recommendations;
• Other texts (AC) through inbound / outbound / reciprocal citation;
• Lists of recommendations and other automatically generated content (PC), through inclusion in it;
• Alternative metrics (PCs), whether primary metrics or complex systems — for example, RG Score or Research Interest.
In this context, it is necessary to focus on the duality of the nature of the text dynamics in the new social conditions. Thus, the "active" nature of non-human communication in ASNS does not contradict, and as a result, does not remove the need for "network work" of researchers, which can be determined by analogy with the practices of SMP (social media promotion).
Research on scientific communication during the late 1960s revealed the extremely high role of informal communication in the processes of disseminating scientific knowledge. In particular, we would like to highlight several elements of those theoretical developments, specifically the concept of "unplanned communication" [Menzel, 1962] and the concept of "translator" [Scientific and Technical Communication, 1969]. Both concepts can be considered as system-forming factors of informal scientific communication, which determine its effectiveness in comparison with formal — mainly intra-organizational, monoparadigmal and publication.
The term "unplanned" scientific communication is understood as spontaneous interaction with scientific information, which can take various forms, from "corridor conversations" with another researcher to an accidentally found book on a table in a library. One way or another, "unplanned" scientific communication provides the researcher with access to information about the existence of which he did not know and, as a result, which he could not detect using a direct search query.
In turn, the concept of "translator" characterizes the communication of two researchers from various scientific fields/disciplines/paradigms/institutions. In some sense, this is the basis of transdisciplinary scientific communication, which implies the transition of various elements of knowledge between scientific disciplines through the communication of diverse specialists, when concepts from one sphere of knowledge, after transformation into the language of another sphere (translation), can be used to solve problems of a completely different science , thereby providing a transfer of scientific concepts and realizing their heuristic potential.
Also, informal scientific communication allows us to overcome a number of limitations inherent in the formal: providing hypotheses and raw-data transfer, as well as overcoming the time lag. Both of these limitations are due, first of all, to the technical capabilities of the "analog" scientific communication provided by scientific journals. Obviously, the format of the scientific article, despite its apparent lack of alternativeness within the framework of the modern neoliberal model of scientific communication, cannot be considered as optimal, which leaves room for some alternative channels — including academic social networks.
The information crisis that determines the dysfunctionality of classical information channels is overcome due to the integration of new channels, ASNS, into the scientific communication, which are an assembly of classical communication practices, quasi-formalized practices of informal scientific communication, as well as auto-determined practices due to the processes of mutual influence of ASNS and researchers at each other.
We call such an assembly a "techno-social system". In this context, it is a special network information management mechanism that operates on the basis of a semi-automated distribution of processes for selecting relevant information between a user, his "Followings" network and ASNS. Thus, the user's "news feed" is formed not only on the basis of his attributes — disciplinary affiliation, scientific interests, expertise and skills, but also on the basis of information from his (relevant) colleagues to whom he is subscribed.
Such a decentralized network forms a common landscape where the information flows are split into relevant groups, and low-quality information is forgotten. This can be represented in the form of a fractal, where individual networks form a group, and then it moves to more and more large-scale levels. Such a techno-social mechanism makes possible the systematization of huge information volumes, thereby overcoming the limitations of technical systems and social mechanisms of information management.
Such interaction is provided by the following elements:
• Archiving of scientific information
◊ Formal (publication) forms of scientific communication:
• Publication attributes
• Texts
• Citation lists
◊ Informal (non-publication) forms of scientific communication:
• "Q&A"
• Commentaries
• Recommendations
• Open reviews
• etc.
• Content Indexing
◊ Statistical indicators (including)
• Primary statistics
• Complex indicators — RG Score h Research Interest;
• Formation of information flows based on:
◊ Automatically aggregated content
◊ "Similar Research"
◊ "Who to follow"
◊ "Questions we think you can answer"
• etc.
◊ Users' content based on "techno-social information system"
• News feed "Home"
• Other automated mechanisms of recommending relevant content
◊ Direct interaction between users
• Public forms
• Private forms
The techno-social interaction forms three types ofinteraction with scientific information within ASNS:
Production of information, including uploading new materials to the network, activities within the Q&A section, commenting, and so on. In general, it consists of all forms of interaction with the network, as a result of which new information is generated — whether it be a scientific publication or just a comment.
Circulation of information provided by automatic aggregators, techno-social system and interaction between users.
Accumulation of information — indexing and long-term storage of information. ASNS accumulate not only user-generated content, but also the vast majority of "digital traces/ fingerprints" — information characterizing the interaction of "active" and "passive" communicators. In other words, the accumulation of information forms an archived level
of user communication, which, along with "publication" content, can be considered as a source of scientific information.
Such an information system ensures the implementation of another principle of scientific communication, developed more than half a century ago — re-issuing information (updating). Thus, an unprecedented level of inclusion of published and non-published information in the structure of actual scientific communication is determined, which significantly expands the information landscape, while providing researchers with effective tools for its social development.
Objections
This study attempts to obtain data on the number of users from each country, data on the representation of each scientific macro discipline and each direction according to RG internal classification, as well as carry an exploration study for a longitudinal analysis on these indicators.
Methods
Our empirical study consists of several stages, two of which we show in this paper. So, our sociological analysis of the academic social network ResearchGate is based on the transition from general to particular, which is caused by the lack of empirical studies that could be considered representative of the entire ReserachGate environment. Understanding that network is a set of interconnected nodes with bottlenecks and, in fact, can be considered a network of networks, we still want to get a comprehensive, global assessment of this space.
Thus, at the first stage, our task is to get two basic characteristics of the field — the geographical and disciplinary distribution of researchers. The details of the required data set may vary depending on research tasks. However, the structure of the social network made it possible to consider several research models for macro-characteristics of the disciplinary distribution of researchers. To determine the macro-characteristics, it would be sufficient to obtain information from the main disciplinary ranking system of the ResearchGate, which includes 24 macrodisciplines. Nevertheless, for further research at the regional level, as well as to detail results, it would be preferable to obtain data on sub-disciplines and specific areas of research work. A similar situation exists with the geographical distribution of users. For macroanalysis, we could use only the distribution by country, which during the analysis could be transformed in accordance with any additional factor. For example, we can consider the distribution by continents, macroregions, economic zones, etc. The distribution by countries seems to us to be the optimal solution, allowing to move both to more global categories and to consider the results taking into account additional regional statistics. Developing the design of the study, we avoided the use of any personal information and, in particular, took all possible steps to avoid violating the confidentiality of private data even in the conditions of publication and as a result of free access to it. We have done it for ethical reasons, since the publication of personal data on the network is for professional communication, and not for social monitoring. Moreover, considering the macrostructure of the network, we could not request the permission to participate in the study from each individual user, so we saw the use of data in a generalized form to be the only ethically correct path. Thus, the required data array can be described as:
1. Determination of the current number of network users
2. Representation of the distribution of users by country
3. Representation of distribution of users by discipline and, preferably, subdisciplines.
4. The data of both distributions must be interconnected, so we must determine the share of each country in each discipline and the disciplinary composition for each country.
5. Array must not include personal data
6. The array must be based on published (public) information that is freely available in order to ensure the possibility of received data verification.
Based on an analysis of relevant research on ASNS, we have identified several hypotheses about the studied distributions, based on regional samples. As a part of our research, we strive to verify them. In particular, we focus on two processes requiring the "global" verification. First of all, we try to test the hypothesis of J.L. Ortega on the homogenization of the social structure of academic social networks [Ortega, 2017]. This hypothesis can be confirmed if there is a trend towards an increase in the growth of researchers from social and humanitarian spheres and at the same time a decrease in the growth rate in the natural and technical sciences. In this context, it is necessary to mention that we are talking about the percentage of already registered users in these disciplines. Since we do not have data characterizing the global number of specialists in each of these areas. In other words, we cannot exclude that there are much more representatives of technical or natural sciences than humanitarian or vice versa. By the way, the search for such data and their comparison with our results seems to us an extremely interesting direction for further research.
We define the second hypothesis under consideration as the "hypothesis of the globalization of social composition". In essence, it implies social processes similar to homogenization, but proceeding at a geographical level. In other words, we assume an imbalance in the geographical dynamics of user growth, in which non-Western countries are currently showing a higher pace compared to the G7 countries. Of course, we take into account that there are probably fewer researchers in these countries.
Obviously, it is impossible to verify both hypotheses "statically"; therefore, the design of our study was built in a longitudinal manner. Therefore, we can highlight another characteristic of data: the possibility of re-collecting. Nevertheless, the Internet research is subject to excessively rapid obsolescence, which may offset the heuristic potential of our work. Checking the trends of homogenization and globalization, from our point of view, requires at least four procedures for collecting and comparing data at least once every three months. It will be optimal to conduct eight such collections. Thus, we decided to divide the report on the empirical part into two separate parts. The first part presents the initial array obtained and the analysis of geographical distribution, as well as the first data collection as part of a longitudinal study, showing the dynamics for six months. The second part focuses precisely on the trends of dynamics and comparison of the main indicators obtained in the first and last collection two years apart. Interim results and materials not included in those two publications will be published in the open diaries of the project in ResearchGate on our personal pages.
We used the GET request method to collect statistics on the number of registered users — one for each country represented in the social network. Totally, 255 requests were processed (the number of countries in the RG system). ResearchGate has 302 disciplines among 255 countries to choose from. When changing the values of discipline and/or country, the quantitative information is updated, while the page as a whole does not reload, which indicates the use of asynchronous data loading. Client data exchange is carried out by the GET-request method, two attributes are transferred to it as parameters: country and
discipline. The Python Requests library was used to collect information. So, it was possible to request statistical data for each of the 255 countries represented in ResearchGate. For our aims the Python version 3.7 was used, including the following packages: json, requests, time, pandas, lxml, urllib.
Results
The first data collection was conducted in October 2019. The total number of network users was 14,849,179 people. It correlates with official figures from press releases. Of course, the press releases show the number of 15 million users, however, the dynamics of the network (allowing for a decrease) and the fact that the difference between these numbers is just over 1%, allow us to believe that the obtained result is true. Moreover, it indirectly confirms the veracity of figures from the company's press releases, so we can use their data without any fear of encountering the negative impact that is possible during the marketing campaign.
In order to initially systematize the information received and ensure the visibility of the graphs, all countries were divided into 11 categories according to the number of users.
Category: Number of users:
0 category 0-100
1 category 101-500
2 category 501-1000
3 category 1001-5000
4 category 5001-10000
5 category 10001-50000
6 category 50001-100000
7 category 100001-250000
8 category 250001-500000
9 category 500001-1000000
10 category 1000001-5000000
Only two countries have overcome the million barrier — the United States (3,232,138 users) and the United Kingdom (1186,199 users). At the same time, the number of users from the USA is 2.7 more than the number of users from the UK. A total of 4,418,337 researchers represent these two countries, representing 29.8% of the total number of ResearchGate users.
The ninth category includes three countries — China (700,112 users), Germany (708,336 users) and India (798,222 users). Thus, the total "volume" of the ninth category amounted to 2206670 people, which is approximately 14.7% of the total number of users.
The eighth category includes Australia (417,802 users), Brazil (413,492 users), Canada (4,58871 users), France (377,798 users), Italy (357,258 users), Netherlands (277,065 users) and Spain (328,329 users). In total, the eighth category exceeds the ninth (2 630 615 users), accounting for about 17.5% of the total number of users.
The 7th category includes 19 countries — Belgium, Colombia, Egypt, Indonesia, Iran, Ireland, Japan, Malaysia, Mexico, Poland, Portugal, Russia, South Africa, South Korea, Sweden, Switzerland, Taiwan, Thailand and Turkey. The average number of users in a group is 159980 people. The maximum value within the group is Indonesia (247,394 users),
the minimum is Ireland (102,204 users). In general, countries of the 7th category include 3039621 users, which is approximately 20% of the total number of users.
In turn, the 6th category also includes 19 countries — Argentina, Austria, Chile, Czech Republic, Denmark, Ecuador, Finland, Greece, Hong Kong, Israel, New Zealand, Nigeria, Norway, Pakistan, Peru, Philippines, Saudi Arabia, Singapore and Vietnam. The average number of users in a group is 79444 people. The maximum value within the group is for Pakistan (98798 users), the minimum for Vietnam (51015 users). In general, countries of the 6th category include 1509432 users, which is approximately 10% of the total number of users, that is, two times less than the 7th category.
Thus, users from countries included in the categories from 6th to 10th make up approximately 92% of the total number (13,804,675 users).
Based on the obtained data, we developed a world map with gradient display of all categories. It can be seen in the multimedia application for this article, using a web link or a QR code.
At the second stage of data processing and visualization, we were faced with the need to compile graphical models of each macro discipline RG. We have chosen the method of proportional areas as the optimal way to visualize the obtained data and finalized this type of visualization, dividing all countries by macro-regions and highlighting each group with its own color. Thus, in addition to visualizing the shares of leading countries in each of the disciplines, we also achieved the ability to visualize their shares by macro-regions. In order to optimize the visualization of the results, all sub-disciplines from the original data array were distributed over 22 macro disciplines. It is necessary to pay attention to the fact that since the sub-disciplines within the framework of the main differentiation system and their analogs within the framework of the studied array are different, we cannot exactly reproduce the "classical" division into 24 disciplines used in ResearchGate. Nevertheless, all the obtained variables were attributed in accordance with the following disciplinary macro groups: Agriculture Science, Biology, Chemistry, Economics, Education, Engineering, Entertainment and Arts, Environmental Science, History, Humanities, Interdisciplinary Life Scientists, IT, Law, Linguistics, Mathematics, Medicine, Physics, Psychology, Religious Studies, Security and Defense, Social Science, Sport Science. All 22 graphics are presented in a multimedia application.
At the third stage we compared the results of the first (October 2019) and second (April 2020) data collection. The number of users increased by 1,525,426 users and amounted to 16,374,605 people, respectively. Our task was to identify the main characteristics of the dynamics of the social composition of the ResearchGate network in order to test the hypotheses of "homogenization" and "globalization" of ASNS.
To test the homogenization hypothesis, we compared the number of users for each macro discipline, measuring its growth. Despite the superior growth rate of researchers from technical and natural sciences in absolute numbers (see Diagram 2), researchers from the social sciences and humanities demonstrate higher rates in percentage terms.
As can be seen from Chart 1, the highest growth rate is demonstrated by the group of Humanities and Entertainment and Arts — by 14.8%. The ten percent threshold was overcome by Law — 10.71%, Linguistics — 10.3%, Religious Studies — 11.14%, Security and Defense — 12.19%, Social sciences — 10.83%, Economics — 11.18%, Education — 10.9% and History — 10.87%. At the same time, the groups most represented on the network: Biology, Medicine and Chemistry — showed an increase of 5.53%, 5.76% and 5.20%, respectively.
Chart 1. ASNS ResearchGate User Growth (by macro disciplines, in percent)
Thus, the increase (in percentage terms) demonstrates a clear tendency to balance and homogenize the social composition of ASNS ResearchGate. This can be caused by both the depletion of non-network resources of the natural and technical sciences, and the reorientation of humanities to the ResearchGate environment. One way or another, the results obtained are intermediate and the nature of the fixed dynamics still requires more deep study.
In order to test the hypothesis of the globalization of academic social networks, we analyzed and visualized the growth of users in each country. Based on the obtained data, we have developed an interactive world map (presented in a multimedia attachment to the article).
Tab. 1 shows data on countries leading in the total number of users and in growth (in absolute numbers).
The second column (Top-20 (total users)) lists the 20 leading countries by the total number of users (in descending order).
The third column (Top-20 (total users / growth rate)) characterizes the distribution of the same countries among themselves in growth. For example, Germany is the fourth country in ResearchGate in terms of the total number of users, as well as the third largest
Religious Studies Security & Defense Humanities History Linguistics Entertainment and Arts Sports Science Law Education Mathematics Psychology Agricultural Science Interdisciplinary Life Scientists Economics Environmental Science Chemistry IT
Social Science Engineering Physics Biology Medicine
1 8455
■ 12792
■ 30266
■i 35237
■■ 34831
mm 62665
mm 48439
wmm 66970
mmmmm 98971
^mmm 67293
50732
121617
36607
2 r- 383476
j- 425754
1000000 2000000 3000000 4000000 5000000 6000000 7000000 8000000 9000000 10000000
Chart 2. The growth of users of ASNS ResearchGate relative to the number of representatives of each disciplinary group (in absolute numbers)
growth in the period from October 2019 to April 2020. The numbers in parentheses show the differences between the position of the country in terms of the total number of users and their growth in absolute numbers. The countries highlighted in color show negative growth dynamics relative to their position in the Top 20 in terms of the number of users. We believe that these countries are approaching the point of saturation and, apparently, are exhausting social resources for growth, however, to test such hypotheses, longitudinal studies and taking into account local specifics are required.
The fifth column (Top-20 (growth)) lists the 20 leading countries in terms of user growth (in absolute numbers). This list is almost identical to the list from the third column, however, it includes Poland, Russia and the Philippines. As you can see from the fourth column, these three countries are not included in the Top 20 by the number of users, however, they show a significant increase in users in absolute numbers. Further study of the social dynamics of the academic social network ResearchGate requires longitudinal research and additional data for a year and a half (at least).
The available data are not yet sufficient to draw any final conclusions. If the trend of rapid, but lagging growth that we have identified continues, in the future the representation of researchers from countries that are not currently leading will grow. The growth rates in the leading ones will obviously decline. However, the growth rate of registered researchers is limited by their total number in each country. Consequently, even when one hundred percent presence of researchers from countries with initially a small number of them is achieved, their indicators will still remain rather small.
The data obtained as part of exploration study of the geographical distribution of users demonstrate significant dynamics in the development of the network (even based on data for six months), so we decided to collect data on geographical and disciplinary distribution every three months throughout 2020 and the first half of 2021.
Top-20 (total number of users) Top-20 total users (growth order distribution) Top-20 (growth) in total number of users distribution Top-20 (growth)
1 USA USA 1 USA
2 UK UK 2 UK
3 India Germany (+1) 4 Germany
4 Germany India (-1) 3 India
5 China China 5 China
6 Canada Canada 6 Canada
7 Australia France (+2) 9 France
8 Brazil Indonesia (+5) 13 Indonesia
9 France Italy (+1) 10 Italy
10 Italy Spain (+1) 11 Spain
11 Spain Turkey (+2) 14 Turkey
12 Netherlands Brazil (-4) 8 Brazil
13 Indonesia Australia (-6) 7 Australia
14 Turkey Netherlands (-2) 12 Netherlands
15 Japan Malaysia (+1) 16 Malaysia
16 Malaysia Mexico (+1) 17 Mexico
17 Mexico Colombia (+2) 19 Colombia
18 Islamic Republic of Iran Japan (-3) 23 Poland
19 Colombia Republic of Korea (+1) 25 Russian Federation
20 Republic of Korea Islamic Republic of Iran (-2) 33 Philippines
Table 1. Leading countries by the total number of users and by their growth
Chart 3 shows the Top 20 countries in terms of user growth (as a percentage) between October 2019 and April 2020. Countries with the number of users less than 10 people by October 2019, were excluded from this list due to very high percentage values in case of an increase even by one or two researchers. It is worth noting that the growth of these countries in absolute numbers varies significantly.
As it can be seen from Table 2, while Nauru and Tuvalu show an increase in users of 4 and 6 users respectively, and due to their low starting positions, they are in the Top 20 countries in terms of growth (in percent), in the list we can also identify countries that really show high dynamics. So the dynamics of the Republic of Uzbekistan is of particular interest — this country demonstrates an increase of 62.7%, which, taking into account its "starting value" of 3305 users, helped to reach the value of its representative office in ResearchGate in 5377 users in six months. Myanmar and Azerbaijan (the similar figures with the Republic of Uzbekistan in October 2019), over the past six months have shown growth dynamics of only 23.4% and 22.3%, respectively, which allowed them to get into the Top 20 countries in relative growth dynamics. We believe that this breakthrough may be caused by the domestic policy of the Republic of Uzbekistan. However, this aspect also
Chart 3. Top 20 countries in terms of user growth (in percent)
Country October 2019 April 2020 Growth
Anguilla 82 160 78
Falkland Islands 25 48 23
Uzbekistan 3 305 5 377 2 072
San Marino 61 97 36
Tuvalu 12 18 6
Chad 129 176 47
Nauru 11 15 4
Turks and Caicos Islands 94 125 31
Northern Mariana Islands 83 107 24
Gibraltar 207 265 58
Marshall Islands 40 51 11
Belize 726 925 199
Democratic Republic of Congo 1 367 1 692 325
Jersey 540 668 128
Myanmar 3 844 4 743 899
South Sudan 159 196 37
Tonga 108 133 25
Timor-Leste 226 277 51
Azerbaijan 3 796 4 643 847
Central African Republic 63 77 14
Table 2. Top 20 countries in terms of user growth (in percent) — absolute numbers
requires more detailed study in the framework of longitudinal research and taking into account local specifics.
Unfortunately, the format of printed journal does not allow to demonstrate all the materials obtained in the course of the study. You can find more of them on the journal website by following the link (http://sst.nw.ru/global-distribution-of-digital-scientific-communication-Case-of-asns-researchgate/) or using the QR code.
In this application there will be:
Visualization of the geographical distribution of Rg users by discipline using the proportional areas method;
A map of the growth rates of ResearchGate users by countries of the world, etc.
Conclusion
In the framework of this article, we began to develop the concept of a theoretical understanding of scientific communication in the context of ASNS. We believe that the consideration of these platforms as an integrated sociotechnical mechanism for the generation, development and accumulation of information flows under conditions of information overload is the most stable theoretical construction. It should be borne in mind that ASNS cannot be considered only as an environment or only as an instrument of scientific communication, since their nature is much more complex. Also, ASNS should not be studied as a space for the implementation of preexisting scientific communication practices only. We are convinced that this relatively new entity in the scientific landscape determined the essential transformations of some practices, and also led to the emergence of new forms of interaction. Obviously, these transformations require further comprehensive study at both the macro and micro levels. Moreover, there is absolutely no doubt that there is a clear inadequacy of the use of quantitative methods only for these purposes — the design of a comprehensive ASNS study should include qualitative methods, as well as realize the potential of both "contact" and non-reactive methods of sociological research.
One way or another, we are only at the very beginning of the path, and it constrains the possible conclusions. So, in addition to continuing the theoretical conceptualization of scientific communication in the context of ASNS, we are faced with the task of conducting a full-fledged longitudinal study. Despite the initial conclusions and hypotheses, a half-year step is absolutely not enough to study the trends in the social dynamics of ASNS. Changing geographic and disciplinary structure requires longer analysis and data collection. Disciplinary distributions are still the most unbalanced aspect of each academic social site as it was shown by J.L. Ortega [Ortega, 2017]. But even at this initial stage we can already talk about some movement of ASNS towards the homogenization of disciplinary and geographical structure, although we are extremely wary of any conclusions.
Thus, further study of the dynamics on a global scale is possible and important for our purposes, however, it will be practically impossible to make any conclusions on its basis without conducting local research with the use of additional statistical sources and observing the scientific policy in each country. We urge researchers of modern scientific communication not to ignore the structural differences of the regions when drawing conclusions based on the global distribution that we are studying. Regional studies, a priori, have a higher heuristic potential studying local groups in ASNS, since local specificity may not be captured by the main network metrics, but nevertheless act as an important factor in scientific communication.
We plan to continue our global monitoring of the dynamics of ASNS ResearchGate, as well as conduct a more detailed study of the social dynamics of researchers from the CIS and BRICS countries. We see these cases as extremely interesting, since we have revealed the anomalous activity of researchers from Uzbekistan and Azerbaijan, and we believe that it may be due to the scientific policies of these countries or some other local
factors. Studying the CIS case will allow us to consider the dynamics of the development of "digital" scientific communication in the territory of the former Soviet Union. In turn, the BRICS case is interesting to us precisely because of the significant geographical spread of the participating countries, the specifics of the scientific systems of each of these countries, as well as the cultural and linguistic differences of these countries, both among themselves and in the framework of the global scientific dialogue.
We believe that the research of scientific communication in ASNS will soon change its design and become multi-factorial, that is, include in the analysis information about the number of scientists in the region, their disciplinary distribution, features of scientific administration and reporting. A network of their contacts may also be an important factor, since global science is only one of the models of scientific interaction, while the model of national or sovereign science probably does not facilitate the entry of researchers into global ASNS. Significant in this context for us are infrastructural studies, namely, questions about the availability and specificity of the local Internet in countries. For example, ASNS can be used to access research texts in countries where access to Google Scholar is difficult. Again, access to WoS, Scopus and other abstract databases and article banks is not provided everywhere. In addition to the technical infrastructure, it is also necessary to take into account the socio-economic, namely, the distribution of English in the region, the fact of buying subscriptions to foreign articles and magazines, especially local laws in matters of science and the Internet regulation, etc.
In the future, from studying the ASNS macrostructure of its dynamics, it will be necessary to proceed to its study at the micro level, namely, strategies, motivations and practices of researchers (fortunately, such studies already exist and, in some cases, they differ in significant results and heuristic potential). Academic social networks are flexible systems that change not only "from above" (that is, by the development team), but also "from below". In other words, user practices are driven by network mechanisms, but the network itself depends on practices and changes in accordance with them. The scientific communication market is changing, which can also act as a factor in the transformation of networks and communication practices. A transformation factor may also be a change in the system of scientific administration, the development of communication technologies, etc. Studying only the macrostructure and basic metrics of the network significantly narrows the range of potential results of such studies, so that out of the focus of researchers may be the most important aspects of the functioning of modern science.
References
Alarcon-del-Amo, M.-C., Gomez-Borja, M.-Â., Lorenzo-Romero C. (2015). Are the Users of Social Networking Sites Homogeneous? A Cross-Cultural Study, Frontiers in Psychology, no. 6, pp. 1-15. https://doi.org/10.3389/fpsyg.2015.01127.
Almousa, O. (2011). Users' Classification and Usage-Pattern Identification in Academic Social Networks, in: Proceedings of 2011 IEEE Jordan Conference on Applied Electrical Engineering and Computing Technologies, pp. 1-6. IEEE. https://doi.org/10.1109/AEECT.2011.6132525.
Barnett, M.P. (1964). The Information Explosion, Nature, no. 4945, p. 585.
Boyd, D., Ellison, N. B. (2007). Social Network Sites: Definition, History and Scholarship, Journal of Computer-Mediated Communication, no. 13, pp. 210-230.
Committee on Scientific and Technical Communication (1969), Scientific and Technical Communication. A Pressing National Problem and Recommendations for its Solution: a Report by the
Committee on Scientific and Technical Communication of the National Academy of Sciences — National Academy of Engineering. National Academy of Sciences.
Dafonte-Gomez, A., Miguez-Gonzalez, M.I., Puentes-Rivera, I. (2015). Academic Social Networks: Presence and Activity in Academia.edu and ResearchGate of Communication Researchers of the Galician Universities, in: Proceedings of 2015 10th Iberian Conference on Information Systems and Technologies (CISTI), pp. 1-6. IEEE. https://doi.org/10.1109/CISTI.2015.7170594.
Elsayed, A. M. (2016). The Use of Academic Social Networks Among Arab Researchers: A Survey, Social Science Computer Review, vol. 34, no. 3, pp. 378-391. https://doi. org/10.1177/0894439315589146.
Fernandez-Marcial, V., Gonzalez-Solar, L. (2005). Research Promotion and Digital Identity: The Case of the Universidade da Coruña, El profesional de la información, vol. 24, no. 5, pp. 656-664.
Green, J.C. (1964). The Information Explosion: Real or Imaginary?, Science, vol. 144, no. 3619, p. 648.
Jordan, K. (2019). From Social Networks to Publishing Platforms: A Review of the History and Scholarship of Academic Social Network Sites, Frontiers in Digital Humanities, no. 6, article 5. https://doi.org/10.3389/fdigh.2019.00005.
Komljenovic, J. (2018). Big Data and New Social Relations in Higher Education: Academia. edu, Google Scholar and ResearchGate, in: R. Gorur, S. Sellar, G. Steiner-Khamsi (eds.), World Yearbook of Education 2019: Comparative Methodology in an Era of Big Data and Global Networks, pp. 148-164. Routledge.
Mason, Sh. (2020). Adoption and Usage of Academic Social Networks: A Japan Case Study, Scientometrics, vol. 122, no. 3, pp. 1751-1767. https://doi.org/10.1007/s11192-020-03345-4.
Menzel, H. (1959). Planned and unplanned scientific communication. Proceedings of the International Conference on Scientific Information, pp. 199-243. National Academy of Sciences.
Mikki, S., Zygmuntowska, M., Gjesdal, 0. L., Al Ruwehy, H. A. (2015). Digital Presence of Norwegian Scholars on Academic Network Sites — Where and Who Are They? PLoS ONE: e0142709, 10 (11). https://doi.org/10.1371/journal.pone.0142709.
Ortega, J. L. (2016). Social Network Sites for Scientists: A Quantitative Survey, Chandos.
Ortega, J. L. (2015). Disciplinary Differences in the Use of Academic Social Networking Sites, Online Information Review, vol. 39, no. 4, pp. 520-536. https://doi.org/10.1108/0IR-03-2015-0093.
Ortega, J. L. (2017). Toward a Homogenization of Academic Social Sites: A Longitudinal Study of Profiles in Academia.edu, Google Scholar Citations and ResearchGate, Online Information Review, vol. 41, no. 3, pp. 812-825. https://doi.org/10.1108/0IR-01-2016-0012.
Thelwall, M., Kousha, K. (2014a). Academia.edu: Social Network or Academic Network?, Journal of the Association for Information Science and Technology, vol. 65, no. 4, pp. 721-731. https:// doi.org/10.1002/asi.23038.
Thelwall, M., Kousha, K. (2014b). ResearchGate: Disseminating, Communicating, and Measuring Scholarship, Journal of the Association for Information Science and Technology, vol. 66, no. 5, pp. 876-879. https://doi.org/10.1002/asi.23236.
Thelwall, M., Kousha, K. (2017). ResearchGate articles: Age, Discipline, Audience Size and Impact, Journal of the Association for Information Science and Technology, vol. 68, no. 2, pp. 468-479. https://doi.org/10.1002/asi.23675.
Van Noorden, R. (2014). Online Collaboration: Scientists and the Social Network, Nature, vol. 512, no. 7513, pp. 126-129. https://doi.org/10.1038/512126a.
Глобальное распределение цифровой научной коммуникации: кейс ASNS ResearchGate
Георгий Александрович Николаенко
магистр социологии, научный сотрудник, Санкт-Петербургский филиал Института истории естествознания и техники им. С.И. Вавилова Российской академии наук; e-mail:eastrise.spb@gmail.com
Роман ВячеславовичМалюшкин
магистр информационных технологий, независимый исследователь Санкт-Петербург, Россия; е-mail: malyushkinr@gmail.com
Анна Викторовна Самокиш
кандидат исторических наук научный сотрудник, Санкт-Петербургский филиал Института истории естествознания и техники им. С.И. Вавилова Российской академии наук; e-mail: tomasina84@mail.ru
В статье описан первый этап комплексного изучения академической социальной сети ResearchGate.net. Задача работы состоит в том, чтобы сформулировать теоретическую основу для исследования сайтов академических социальных сетей (ASNS) и проанализировать первые данные о глобальном географическом и дисциплинарном распределении ASNS ResearchGate.
Теоретическая концепция предполагает анализ академических социальных сетей как сложных социально-технических систем для производства, распространения и накопления научной информации. Следовательно, академические социальные сети рассматриваются одновременно как среда и инструмент научного общения. В качестве среды ASNS определяются как квазивиртуальная область научного взаимодействия, в которой различаются «активные» и «пассивные» типы коммуникаторов. В свою очередь, ASNS как инструмент представляет собой комплексную систему для архивирования научной информации, индексации контента и обеспечения информационных потоков. Особое внимание уделено соотношению формальной и неформальной научной коммуникации в контексте информационного перенасыщения, которое определяет новые виды практик. Акцент сделан на макрохарактеристиках платформы — распределении всех пользователей по странам и научным дисциплинам. Полученные данные визуализируются с помощью диаграмм, выполненных методом пропорциональных площадей, графиков и карт. В статье представлены результаты пилотажного исследования, позволяющего изучить социальную динамику академической социальной сети ResearchGate, в том числе тенденции дисциплинарной гомогенизации и социальной глобализации на этом сайте.
Полученные результаты дают наиболее широкую картину текущей социальной структуры ResearchGate. Характеристика всего массива пользователей, в отличие от использования локальных выборок, позволяет преодолеть ряд ограничений, связанных с репрезентативностью. Разработанный дизайн исследования позволил зафиксировать полугодовую глобальную динамику социальной структуры сети, что позволяет нам проверять локальные гипотезы по всей платформе. Мы также считаем, что такой дизайн исследования может поставить ряд новых вопросов и дать результаты, которые увеличат эвристический потенциал изучения научной коммуникации в Интернете.
Ключевые слова: ASNS, ResearchGate, социальные сети, академическая коммуникация, социальная структура, статистика, вебметрики, альтметрики, академические социальные сети, глобализация.
