6.1.1 Profiles, the Main Asset of a Network

The principal asset of a social networking site is its users because without them there is simply no site. As a result of this, it is not surprising that the owners of social sites are proud of the number of users that they host and celebrate when they reach an important milestone. The number of users registered in a social site is always considered an indicator of size and describes the power of a site to attract and accumulate members. A highly populated network therefore favours networking among partners, the creation of collaborative structures and the provision of new relevant content. To some extent, each social site needs a critical mass to develop these basic functionalities and begin to attract more potential users, in turn increasing its value. Following Reed’s law, the value of a social site increases as more members join that network because the number of possible new groups or connections exponentially increases (Reed, 2001).

Figure 6.1 presents the total number of users registered in each social site in 2015. The site with the greatest number of users is Academia.edu, with 23 million scholars, followed far behind by ResearchGate with 7 million and Mendeley with 4 million. The smallest networks are Zotero (130,000), CiteUlike (208,000) and UniPHY (300,000). In the case of BiomedExperts it is necessary to specify that only 471,000 profiles were validated, and in BibSonomy only a much a reduced group of users (9,000) remain active. These figures illustrate that, in the panorama of the academic social sites, document sharing systems are consolidating, being the platforms that attract most users, far and above other models such as reference management and social bookmarking sites. Of these, Mendeley is the only non-document sharing platform that may be considered a successful site with 4 million users. The most surprising is that Academia.edu has four times the number of profiles of its nearest competitor. This disproportionate number of users could be due to the fact that many of these profiles may be non-academic independent members that are using the site only for exploring full-text documents instead for uploading items.

However, the number of users in a site has to be observed from a temporal point of view in order to properly describe the energy with which these sites are growing and projecting their future positions. The annual growth rate was calculated to measure the growing trend of each site. To calculate this indicator longitudinal data were necessary, so it could not be computed for Nature Network, BiomedExperts, UniPHY (no longer available) and BibSonomy (without historical data). However, according to this indicator, the sites that present the greatest increase are the two major document sharing sites, ResearchGate (128 per cent) and Academia.edu (118 per cent), followed far behind again by CiteULike (84 per cent) and Zotero (76 per cent). This result confirms the emerging behaviour of the document sharing sites and the standstill of the other platforms. It is also remarkable that Mendeley (63 per cent) is the site with the lowest growth. For more precise figures an annual growth rate was computed only for the last five years, because the older sites have had more time to accumulate members, whereas the younger ones have not yet developed their potential. Now, Academia.edu (74 per cent) presents the highest growth followed by ResearchGate (50 per cent), while Mendeley (23 per cent) surpasses Zotero (20 per cent) and CiteUlike (10 per cent), the oldest site. This confirms the rise of Academia.edu, Researchgate and Mendeley as the three most popular platforms for social networking for scholars and suggests that the other social instruments are losing the interest of the academic community.

6.1.1.1 Disciplinary Distribution

The thematic distribution of profiles is a key factor to observe if the population of a site is balanced or, on the contrary, the network tends to specialize in a specific research area. The specialization could be positive if the site was designed only for the target users, but may be counter-productive for most sites because this reduces the diversity and therefore limits the future growth to see off potential users from different disciplines. Thus if a network does not present a sufficient range of content and there are not enough colleagues in each subject area, then it is possible that those sites will lose attractiveness to upcoming potential users.

Only five academic sites could be classified thematically, the rest corresponding to specialist sites (i.e. BiomedExperts specializing in Biomedicine and UniPHY in Physics) or platforms that do not arrange their users thematically (BibSonomy). Figure 6.2 shows the distribution of profiles according to four main research areas: Life Sciences, Social Sciences and Humanities, Physical Sciences and Health Sciences. This classification scheme is the one employed by Elsevier in many of their products (i.e. ScienceDirect, Scopus) and is used here because it is clear and helps comparison across platforms. Academia.edu and Zotero show an elevated degree of specialization, with 70 per cent and 61 per cent of profiles belonging to Social Sciences and Humanities respectively. As has been said, the presence of large numbers of humanities scholars and social scientists could pose a risk for the upcoming development agenda, because this fact could limit the attraction of new scholars from other disciplines. On the other hand, Mendeley, ResearchGate and CiteULike present well-balanced distributions with an increased degree of diversification. Even so, it is interesting to note the significant presence of Health Sciences (23 per cent) in ResearchGate and the strong presence of life scientists (37 per cent) in Nature Network (46 per cent). In this same site, Business/Investment (93 per cent), the category of Social Sciences and Humanities, was removed because it could be used as the default class and distorted the results. Due to this, this research area was not graphed for the Nature Network case.

6.1.1.2 Country Distribution

The extension of an academic social site across countries could be an indicator of its success in the research community, because as a site grows, it is more likely to attract new users from different places. Thus a service with a large proportion of international users demonstrates that the platform has reached across the globe and become a model for scholarly networking sites. However, a service with a high proportion of local users is evidence that the site has not taken off and has became in a local tool with few prospects.

In this analysis, only BiomedExperts and UniPHY were excluded because they have disappeared and it was not possible to count the nationality of their users. Table 6.1 presents the percentages of members from their countries in each social site. These are the countries that most users have in every platform. The results show in general that the United States and the United Kingdom are the countries with the highest presence in these platforms, followed by other western and/or Commonwealth countries. Only in exceptional cases do different countries emerge with better percentages, as is case of BibSonomy, where 41 per cent of the users come from Germany. In any event, this important presence of western countries is also due to these countries having large populations of researchers and it is inevitable that they will occupy the top positions.

Table 6.1

Percentage of users by country in each social site and Country Spreading index

Country	ResearchGate	Academia.edu	Mendeley	Zotero	CiteULike	Bibsonomy	Nature Network
United States	22.8	17.1	22.5	43.6	27.2	7.3	32.2
United Kingdom	7.6	8.9	9.7	6.2	10.7	4.0	15.2
Germany	4.6	2.0	6.2	2.7	4.6	41.1	3.7
France	3	2.1	2.7	3.5	4.0	4.6	1.5
Spain	2.7	2.2	4.8	1.9	3.4	2.0	1.2
Canada	3.1	2.4	3.7	3.8	3.3	1.0	2.8
India	5.9	3.6	2.4	2.1	4.8	3.0	11.4
Italy	3	2.9	2.2	0.7	3.3	3.0	1.6
Australia	2.8	2.5	2.2	3.2	3.0	1.0	1.8
CS index	55.9	53.7	63.3	69.3	68.1	73.8	73.6

Next, Figure 6.3 graphs the degree of penetration of each service into a country. This relative index reduces the size effect of large countries with high academic populations. In general, the picture presents a slightly confusing view with a very varied distribution of penetration so it could be said that each social site follows a different pattern. However, with the exception of BibSonomy, the United Kingdom stands out as the site with the best penetration across the board, suggesting that researchers from this country show a positive inclination towards these platforms. This could be due to the fact that many of these platforms are located in this country such as Mendeley, CiteULike and Nature Network. The remaining countries present a varied penetration and the aforementioned case of BibSonomy should be noted, along with the diffusion of Nature Network by Commonwealth countries, in particular India, and the case of Zotero with a majority presence of American users.

Another way to measure the spreading of a platform around the world is to measure the concentration degree of their users in many or few countries. Country Spreading (CS) measures how highly geographically concentrated/spread the population is of a social site. High values of CS therefore describe very local sites that have expanded over just a small number of countries, while low rates of CS correspond to sites that have reached large parts of the world. Academia.edu (53.7 per cent) and ResearchGate (55.9 per cent) are the services that have spread the most, suggesting that these platforms have expanded the most all over the world. On the other hand, BibSonomy (73.8 per cent) and Nature Network (73.6 per cent) are highly concentrated in a greatly reduced space. In the case of BibSonomy this concentration is around German-speaking countries, while Nature Network was mainly instituted in Commonwealth countries. This index could be interpreted as a health indicator because spaces with a wider geographical diffusion are sites with a great development and positive growth as well, as with ResearchGate and Academia.edu. On the contrary, sites with a small geographical spread have disappeared (Nature Network) or remain marginalized to a local environment (BibSonomy, Zotero).

6.1.1.3 Academic Statuses

Academic position could be a proxy for the age of the profiles, at least with regard to ‘academic’ age, because it is common that students would be younger while professors and researchers are likely to be older. It is also true that scholars have different needs and develop different activities according to the academic stage in which they are. As a result of this, the analysis of the academic positions allows us to observe the difference in preferences with regard to the use of one or other type of platform.

Figure 6.4 presents the distribution of users by their academic positions in four academic sites: Academia.edu, ResearchGate, CiteULike and Mendeley. The remaining platforms do not include this information in their profiles. The classification scheme has been developed ad hoc and is expected to cover the great diversity of positions that there are in the academic world. Both document sharing sites, Academia.edu and ResearchGate, show a varied range of academic positions, with a significant presence of senior posts such as Professor, Researcher and Professional. However, CiteULike and Mendeley, two sites specializing in reference sharing, present a disproportionate presence of young scholars such as Student and PhD Student. These opposing distributions could be the result of the different solutions that these platforms offer and the type of users interested in those functionalities. Thus, for example, reference managers such as Mendeley and bookmarking sites such as CiteULike attract many students that need to gather and manage long and exhaustive bibliographies on their research fields to support the start of their scientific career. On the other hand, document sharing sites attract the attention of a wider range of scholars, both younger and older, whose aim is to distribute their academic publications and track their impact in an online environment. In summary, document sharing sites are spaces suitable for any type of scholar, while bookmarking and reference management sites are being used mainly by young students. This, perhaps then, is another cause of the success of document sharing sites over other platforms because they are not limited to a particular age, while other platforms are spaces that principally attract younger scholars.

6.1.2 Publications, Fuelling the Network

If the users are the principal asset, the amount of content that these users contribute to the system is the main wealth of a social site because they represent the results of the activity performed in the network. From an academic point of view, the content of these sites has a very special significance because it transforms these platforms into information resources for searching relevant articles as well as for keeping up to date on the recent literature. Moreover, the social activity of users tagging, reading, viewing and downloading items may be used as an indicator of the quality of these materials, which allows the selection and filtering of documents according to their impact in the online community.

Figure 6.5 shows the distribution of references and documents posted to each academic social site. Nature Network is not included because it did not contain publications, nor is UniPHY because the number of publications hosted on the site is unknown. Mendeley is the site with highest number of references, approximately 114 million being aggregated from all the users’ libraries. However, many of these are duplicated and are not publicly accessible from the Mendeley website. Thus the figure closest to the Mendeley public catalogue is 14.1 million records. In spite of this drastic reduction, this site remains in second place in terms of the number of bibliographic references, while ResearchGate is at the top with 80 million references, of which 19 million also have a full-text copy. The remaining services are far behind, with values under 20 million. Hence, BiomedExperts contained 18 million documents harvested directly from PubMed. Next, CiteULike hosts 8 million citations, while Academia.edu contains 6.1 million references and 4 million full-text documents. Finally, BibSonomy with 3.5 million and Zotero with 2.5 million are products that contain the lowest number of materials which confirms their reduced scope of activity.

The reason for the outstanding performance of Mendeley and ResearchGate could be the fact that their databases are not only fed directly by their users, but that both sites extract bibliographic information from secondary sources. In the case of Mendeley, references mainly come from PubMed and Scopus, while ResearchGate takes its metadata from open repositories. It is also necessary to note that the possibility of uploading a full-text document in a sharing site is much more limited than posting a reference in a reference manager or social bookmarking site, so data on ResearchGate and Academia.edu have to be even better acknowledged. For example, ResearchGate, with its 19 million full-text copies, is now the top supplier of full-text documents to Google Scholar, which shows the transformational power this tool has in the scholarly world (Jamali & Nabavi, 2015).

As with users, the content production rate is also an indicator of the growth of the network and points to the increasing health of the site with the addition of new documents. Annual growth rate was calculated in only four sites: ResearchGate, CiteULike, Academia.edu and BibSonomy, the remaining sites not containing historic data that would facilitate the computation. The site that experienced the highest annual growth is ResearchGate, with an exceptional 320 per cent. However, this was only calculated from full-text documents during the 2012–15 period, and would possibly be a much lower percentage for all documents and the full life period. Next, Academia.edu describes a growth of 43 per cent, followed by BibSonomy (27 per cent) and CiteUlike (20 per cent). Nevertheless, if only the last five years are considered, the growth of BibSonomy (13 per cent) and CiteUlike (7 per cent) is clearly reduced, which confirms again the decline of the social bookmarking sites. Overall, the results show that the services which supply most content (Mendeley, ResearchGate and BiomedExperts) are those that are fed from other secondary sources. It goes to show that the determination of the users is not enough to maintain an up to date and complete information system, and demonstrates that a collaborative environment is not a suitable model for creating acceptable bibliographic resources.

6.1.2.1 Disciplinary Distribution

Another important aspect of the publications referenced or deposited in social sites is the thematic distribution on each social site. As information sources, these spaces are used to obtain bibliographical information and knowing a site’s disciplinary coverage helps in the choice of which platform to use to find relevant documents. Also, the thematic distribution of a social site is of interest in order to decide whether to participate or not in the site because users tend to take part in spaces where they are familiar with the content that the service offers.

Only four sites, the two reference management sites and the two sharing sites, thematically arrange their publications: Academia.edu, ResearchGate, Mendeley and Zotero. The remaining platforms have not been studied for several reasons: they do not contain any publications (Nature Network), they are specialized sites (BiomedExperts, UniPHY) or they do not use any classification scheme as with the social bookmarking sites (CiteULike, BibSonomy). As with profiles, publications were classified according to the Elsevier scheme. Academia.edu (74 per cent) and Zotero (61 per cent) show a clear bias toward Social Sciences and Humanities, because the presence of humanities scholars and social scientists is very high to other academic social sites (Figure 6.6). The other research areas show very low percentages, with the exception of Physical Sciences that is around 20 per cent in both sites. On the other hand, Mendeley and ResearchGate describe opposite patterns. The distribution of publications is more balanced, predominantly Physical Science with 37 per cent and 38 per cent documents and Life Sciences with 24 per cent and 27 per cent respectively. ResearchGate is the site that presents most biomedical content with 25 per cent of the papers, a percentage near to the proportion of physicians registered in the site. As was said in the Profiles section, Academia.edu and Zotero could be considered specialized networks because the great majority of content corresponds to Humanities and Social Sciences, while Mendeley and ResearchGate are balanced platforms with a similar thematic distribution comparable to other scientific information resources (i.e. citation indexes, scientific databases, search engines). The equilibrium in these two sites could be related to the fact that both sites take many of their publications from external sources. It is possible that this practice provides them with stability of content and therefore they are more attractive to a wider range of scholars. Otherwise, on comparing the distribution of profiles and publications, it is interesting to observe that the presence of Social Sciences and Humanities profiles in ResearchGate and Mendeley is higher than the same proportion for publications. This could indicate that this type of user produces less content than scientists from other disciplines.

6.2.1 Groups, the Collaborative Tool

Up to this point, the activity of academic social sites has been explored from an individual view, analysing how each user posts and follows other users. Now, the way in which users group together to produce some action is studied. Groups can be considered the main collaboration instrument of an academic social site because under this instrument different users can carry out tasks together with other members. In this way, this space allows us to observe to what extent users from a social site employ the service more for collaboration activities than for individual purposes.

This instrument is not available in all the academic sites studied. The document sharing services ResearchGate and Academia.edu do not have any similar function, while BiomedExperts and UniPHY also lack of this element. Figure 6.9 shows the number of groups created in each site, in addition to the annual growth in items posted and the percentage of items posted to groups since 2014 (since 2012 in Nature Network). Mendeley is the site that has the most public groups (97,857), followed by Zotero (31,783) and CiteULike (5,085). The large number of groups in Mendeley – three times the number in Zotero – could be due to the difference between them in the number of users. It seems that reference management sites are more successful in the creation of groups than bookmarking sites. Thus the growth rate of new content in Mendeley (112 per cent) and Zotero (94 per cent) is much higher than in CiteULike (42 per cent) and BibSonomy (42 per cent), pointing out the standstill reached by social bookmarking sites and the strength of reference managers as collaborative spaces. The percentage of items posted since 2014 is a measure of group updating as the more posts are located after that period the more current are their content. Thus, BibSonomy (46 per cent) has the most updated groups, followed by Mendeley (38 per cent) and Zotero (35 per cent). As has been seen, BibSonomy is supported by a small number of users that generate a huge amount of activity. This energy is also transferred to groups, as users are utilizing groups as the main networking space.

Another aspect to evaluate in groups is their activity in the context of the entire site, as a means to analysing the weight of these collaborative instruments in the entire activity of the service. Figure 6.10 shows the activity of these groups, calculated as the ratio of the number of posts in each group to the number of members affiliated to this same group. In addition, the percentage of users that decide to join to a group as well as the percentage of items posted to groups in relation to the entire site are calculated. BibSonomy (189.8) again shows by far the highest activity due to the reason given above – a lot of activity is being produced by a small number of users. Zotero (40.1) and CiteULike (32.6) also describe good activity rates but with five times lower activity than BibSonomy. In relation to the proportion of users attracted to groups, Zotero (82 per cent) is the platform in which most users are linked to groups, followed by BibSonomy (18 per cent) and CiteULike (14 per cent). The excessive proportion in Zotero could be due to the fact that most of the users are signed up to this service almost exclusively to participate in groups. However, Mendeley (31 per cent) is the site with the most content supplied to groups, followed by BibSonomy (10 per cent) and CiteULike (8 per cent). One reason for this could be that Mendeley has the largest bibliographic database, which helps the posting of items to groups.

In general, reference management sites are employing groups more profusely than the other platforms, creating a larger number of groups and a greater increase in their activity. This can also be seen in the weight that groups have in the entire site, with a higher percentage of publications and members involved in these collaborative tools. This has come about, firstly by the higher collaborative possibilities of the reference management sites which make possible the generation of bibliographies and the addition of references to personal libraries, and secondly by the gradual stagnation of social bookmarking sites in which activity is going down and limited to small communities such as BibSonomy.