Keywords

Web contents; scholarly reputation; almetrics; academic social networking sites; Science 2.0

7.1 Content as the Underlying Factor of Social Network Development

The first conclusion is that the success or decline of a platform is intimately related to the type of content that it manages. Document sharing systems are positioned as the most successful spaces both for content generation and social networking, in addition to which this type of scholarly social networking site displays better future prospects. On the other hand, social bookmarking sites present an outdated model with a general drop in users, documents and activity that reflects the stagnation of the folksonomy as a paradigm for the creation of social networking sites. Reference management sites, considered to be a more sophisticated version of bookmarking sites, did not take off as social platforms although they remain active as a desktop application for private purposes. It is possible to claim that the achievement of a platform is strongly related to the type of services that it offers and the information or content that it manages. Thus, the sharing of full-text documents is the solution that attracts most scholars as this is the essential activity that researchers have always done in order to be up to date and connected with the most recent and current advances in their fields (Vickery, 2000). On the other hand, the sharing of references is only important for bibliographic tasks and information searching which, as has been seen, only attracts the attention of students and young scholars. Content thus emerges as the key element in the working of these platforms and determines the most successful architecture. It is not surprising that designs such as Nature Network, BiomedExperts and UniPHY have disappeared, as their models did not take into account the exchange and production of content. In the case of Nature Network the networking system was basically based on forums and discussion groups, while BiomedExperts and UniPHY simply did not include any instrument for sharing information among their profiles. With regard to the social bookmarking sites (CiteULike and BibSonomy) and reference managers (Mendeley and Zotero), bibliographic reference is the main information unit that these platforms put in circulation. In these sites, users act as collaborative collectors that interchange and post bibliographic citations with the aim of creating self-organized information systems. In the case of document sharing services (Academia.edu and ResearchGate), documents are now the key materials, shared by the authors themselves through the uploading of full-text copies to their respective profiles. It could therefore be possible to think that there is an evolving change in the nature of these academic sites, in which the content has played a central role in the adaptation and transformation of these models. Thus, from a initial uncertain approach in which it was expected that users would only be put in contact through forums and groups, an active model was developed in which tagging and posting references was the force that shaped the network. This was finally finished with a framework focused on the establishment of networking ties from the exchange of the creations of the users themselves.

7.2 Activity Defines the Typology of Users and Site Usage

The second conclusion is that participation in these platforms is diverse and influenced by the type of service and users. The definition of participation may be blurred and depends on the platform type, the activities that can be developed there and the amount of content that a user could post. Nielsen (2006) explains the 90-9-1 rule which states that 90 per cent of users do not participate in anything, 9 per cent contribute a little and 1 per cent make most of the contributions. However, this rule is not exact and the variations could be used as an indicator of activity and inequality. Nentwich and König (2014) count five types of users according to their degree of participation and suppose that those with the lowest activity are the most frequent users. In our study, if participation is considered to be the number of users that post something to the network, the empirical results show that, in the best-case scenarios, participation is 68 per cent (ResearchGate) and 46 per cent (Academia.edu), while in the worst it is 14 per cent (Mendeley) and 0.9 per cent (BibSonomy). In any case, the results show values higher than estimated by Nielsen’s rule which suggests that the activity in academic social sites is higher than in other web spaces. Moreover, the percentages are very different among them and demonstrate that the participation levels are not the same across platforms, but that the type of activity that these sites develop could be the main influence on the involvement of members. Document sharing services present by far much higher activity than the other services, suggesting that the exchange of documents is a more attractive activity for users than posting references or tagging bookmarks. However, this view if participation is rather flat and only considers whether the user posts or does not post content. The analysis of these spaces has brought us to observe that the degree of participation is much more complex and that users’ activity could be expressed in very different ways. Thus, different typologies of users emerge to develop different activities according their needs and the possibilities of the service. In the case of Academia.edu, this fact was clearly observed when two types of users were detected: viewers make up the majority many of whom are independent users that mainly follow the research activity of other users but scarcely put in content of their own. On the contrary, producers form a smaller group of highly productive researchers that post a large amount of items but they are not interested in following the activity of the network. To some extent, these different types of users arise from the different needs that they have and how they can satisfy them in the platform. In Mendeley, for example, there are four million users with poor networking rates and one of the lowest proportions of users posting content (14 per cent), joining groups (6 per cent) or following members (33 per cent). In this case, users perceive Mendeley more as a private desktop application than as a social networking place. Following this idea, it is possible that many of the passive users observed by Nielsen (2006) and Nentwich and König (2014) would correspond to members that employ these spaces as information resources wherein they explore the site looking for new and up-to-date content, instead of posting items, setting up collaboration groups or responding to questions. This leads us to bear in mind that participation in these spaces consumes a significant amount of time, which would explain why most users have just a basic presence.

7.3 Diogenes Club? Private Over Public Interests

The third conclusion is that users move toward scholarly social networks more for egotistical motives than for collaborative purposes. In a world in which participation in academic social networks has been lauded as an example of the cooperative nature of Science (Calhoun, 2014; Cann, Dimitriou, & Hooley, 2011; Zaugg, West, Tateishi, & Randall, 2011), and where these spaces have been associated with the altruistic spirit of the scholarly community, it is surprising to state that these networking platforms are in the main not being used for collaborative purposes but rather for personal matters. Although we talk on social networking sites, many of these platforms allow the use of these spaces for individual purposes without the need to establish contacts or enter into collaborations with other users. In other words, it could be said that these sites are not so much a meeting room where users chat and develop activities together, but are rather personal spaces where individual actions can be taken advantage of by other members. The reason for the small fraction of ‘active’ users could be due to the remaining members using the network for private purposes, viewing the information that others post, witnessing the discussions that others initiate and tracking the statistics that their publications produce. Social bookmarking sites, the best examples of collaborative space, in truth, are a collection of private libraries made public in which the rest of the users can reuse the work of others. Even then, only 26 per cent of CiteULike’s and 10 per cent of BibSonomy’s members have posted something in their libraries. The remaining members could be using the libraries of others to be up-to-date on specific issues and/or as a specialized information resource. Seen in other way, a few users put a lot of content in a network so that the great majority can capture the most beneficial items for them. Groups, a service precisely oriented to the collaboration between users (Oh & Jeng, 2011), is the clearest example of the poor interest of this community in taking part in joint activities. Hence, the percentage of users enrolled in groups is 13.9 per cent in CiteULike and 18 per cent in BibSonomy, while the average number of posts to groups is lower than the same average to the entire network in CiteULike (18.6 in groups; 25.7 in the network).

The situation with reference management sites is not much better. Mendeley and Zotero were created as desktop applications for individual needs and their jump to the social networking environment has not been as good as expected. In their regard, the system is only able to make public the references that their users want. In Mendeley, the ‘public catalogue’ only contains 12 per cent of all the items posted to the service, showing that their users are not willing to share their references. This pattern is also repeated in Zotero where only 4 per cent of users’ libraries are public. Nor have the networking spaces of these services taken off sufficiently either. Mendeley’s network only has an average of 1 follower per user, while 14 per cent of their users have scarcely added any papers to their profiles. Zotero describes similar networking figures with 11 per cent of followers and 7 per cent of followings. These data confirm the poor networking activity of these platforms and, as in social bookmarking sites, it appears that most of their users only want to manage their own libraries and exploit the public catalogue to capture references. The use of groups is also similar to social bookmarking sites. In Mendeley, only 8 per cent of users are joined to groups and the post average in these services (16.7) is lower than in the entire network (24.9). These results do nothing but cast doubts on the utility of a social networking site in products that mainly offer one service – references management – for purely private purposes.

The situation with document sharing sites is a little different. In these spaces, the main intention is to share publications between their members, which means that the participation levels are higher. Thus, 68 per cent of ResearchGate’s users and 46 per cent of Academia.edu’s post some items to the platform. However, while these rates do not reveal an altruistic behaviour, they could be motivated by a conceited need to monitor the use and impact of their publications in the academic world. The range of metrics that describe the performance of papers, profiles and organizations, and the dashboards that track the usage of the users’ publications, are instruments that encourage the publishing of results, but do not promote collaboration at all. According to networking metrics, these spaces are also the most active with the highest average of followers and followings. However, these activity levels could be not motivated by social or connection needs, but they might be used for searching information. Thus, as we have seen in Academia.edu, many of the users that principally follow profiles are also members who provide little content. In this sense, followers could be users that focus their activity on seeking information through the updates they receive from their followings. On the contrary, producers, who post many publications but follow few people, employ the application mainly for disseminating their output. In this way, both types of users might not be centred in collaboration activities, but they could be looking to meet their personal needs. Questions & Answers is the only collaboration tool used in this type of social networking place, which provides evidence of the limited interest that document sharing sites have in the collaboration process. This function reveals that only 1 per cent of users launch a query and just one in seven follows the replies to these queries, showing again the poor attention to participative environments in the academic community.

To summarize, academic participation in these virtual spaces is more like a kind of Diogenes Club¹ where all the members come together to do private and personal things, while freely taking advantage of the work of their colleagues. To some extent, the academic contacting network is based on a rewards system in which an elite of highly prestigious producers use these spaces to spread their productions in return for watching the impact that these materials are provoking in a huge community of young scholars and academics from developing countries who are following this production so they can be up to date in specific fields.

7.4 Metrics for Managing Reputation and Socializing, not for Research Impact

The fourth conclusion is that social sites for scholars cannot be a substitute, at the present time, for the current scholarly reward system, not even as an alternative. The reasons for this statement include a wide variety of matters. The first is the previous need to define the sense of these indicators and their implications for the research assessment of individuals, papers and organizations. It is easy to précis the meaning of citations, publications, etc., as proxies of scholarly production and impact, but metrics such as views, readers and followers come from very different actions and the significance of these indicators for the evaluation of research is not still clear (Sugimoto, 2015). Perhaps more research studies are needed to uncover the real meaning of these metrics and see to what extent they reflect the type of impact they have and whether, in the end, this is related to academic excellence and prestige. At the moment, many studies have only found average correlations among these measurements caused, in most of the cases, by multiple and uncontrolled interactions (Li, Thelwall, & Giustini, 2012; Ortega, 2015; Thelwall & Kousha, 2014, 2015). Another limitation is that these online networking spaces are closed and their usage, networking and impact metrics are calculated on activity inside the network. This means that similar metrics such as followers/followings, views, downloads, etc., could express different results across platforms so that, for example, users with a high rate of views in one place could go unnoticed in another site. This is explained not only by the intrinsic technical differences in extracting and computing these indicators or simply by the number of users, but also by the important geographical and thematic biases that these populations present. Yet while there are sites mainly occupied by researchers from the same country (Bibsonomy, Zotero), others are unbalanced toward certain disciplines (Academia.edu) and in other cases students surpass the population (Mendeley). These biases could influence these indicators in such way that would make it almost impossible to benchmark between metrics from different services. Another problem not adequately taken into account is the length of time a user or paper is included in a social platform (Thelwall & Kousha, 2014). Networking (i.e. followers) and usage (i.e. views) indicators are also time-dependent metrics, increasing their value according to the length of time that a profile is active on a particular site and which could be different had the profile been on a different site. In other words, the metrics of one user who has been registered in a service for a length of time would be higher than if the author had recently signed up. Another important fact is that signing up to these services is voluntary and the adoption rate by this part of the research community is generally low and different across sites (Haustein et al., 2014; Mas-Bleda, Thelwall, Kousha, & Aguillo, 2014). This means that it is difficult to evaluate some authors while the global view on organizations and research disciplines could be incomplete and limited. Despite these limitations, however, these reputation metrics do offer additional information on the influence of an author and the usage of their research outputs on their online scientific network, describing a different and social dimension to the place that a researcher occupies in the scholarly community. These measurements therefore must also be appreciated but only as signs of networking, popularization and socializing.

7.5 Toward a Societal Commitment of the Scientist to the Citizenship

The fifth and last conclusion is that scholars should get involved in these spaces because of their commitment to the citizens in general and to being transparent to society. In general, one of the changes that the Web brought was that people searched for scientific information with more assiduity and interest, which entailed a higher perception of science and a closer understanding of the scholarly activity on the part of society (Ripberger, 2011). However, it also brought about more public judgement of the research activity as well as increasing the impact of scientific results on daily life (i.e. Medicine, Environmental Science, Economics, Information Technologies, etc.) (Leach, Scoones, & Wynne, 2005). For example, many of the independent users of Academia.edu may not be academics looking for updates to issues in which they are interested. Citizens thus have the right to demand responsibility and results from scientists, not only because the last addressee of the research outputs is society itself, but also because many of the researchers are civil servants and their research investigations are funded by public resources (Hagendijk & Irwin, 2006). In this way, researchers have a public commitment to society to which they have to answer (Molloy, 2011). Hence social networking places are ideal locations where scholars should make public their academic outputs so that society can appreciate the importance of their contributions and be sure of the efforts these professionals make (Martorell & Canet, 2013). Usage (views, downloads) and networking (followers) metrics are good indicators that fit in with the context of the popularization of science and societal impact. This transparent attitude could be an important way to justify the enormous budgets allocated to R&D, and to move the citizenship closer to academic culture. In other words, organizations and countries should promote the use of academic social sites as a way to persuade society of the importance to the country’s health of funding science. Thus a society aware of the efforts of its scientists is a society that will better understand the problems and needs of the scholarly community.

Note