Introduction
Abstract
This first chapter begins with an introduction to the past and current context of social networks for scientists and a review of the rise of the Web and its impact on scholarly communication. It also explores the appearance of the Web 2.0 philosophy and the transformation of the publishing system as a result of the open access movement. Social networking sites are described as predecessors of the academic platforms and the revolution of altmetric indicators is discussed in detail. An examination of the definitions and typologies of academic social sites is then provided as well as a consideration of the business models of the platforms studied. Finally, several methodological aspects are described such as the proposed indicators, and the sources and data extraction processes are explained.
Keywords
Web 2.0; altmetrics; social networks; open access; social networks for scientists
1.1 The Web and the Web 2.0 Concept
At the dawn of the new century the Web had become consolidated into all aspects of life, this new and revolutionary information technology has transformed the information habits of the entire world and had made possible the immediate diffusion of content to any part of the globe. Born into the academic environment, the Net soon reached every facet of human activity, turning information into an important transformational asset for the rising knowledge society (Castells, 2010). The successful changes brought about by this disruptive technology do not simply rely on instant access to an enormous amount of data, videos, pictures, etc., but on the possibility of avoiding those mediators that, at that time, controlled the information flows. This singular characteristic meant that users would became publishers and distributors of their own creations, without any gatekeeper to censure or take advantage of its position (Van Dijck, 2009). Authors that published their own books online, little shops that offered their products direct to the consumer and businesses that advertised themselves on fancy web pages proliferated exponentially. Thus a great part of the information that ran through the Web was content created by its own users, being at the same time sources and receivers according to communication theory (Morris & Ogan, 1996).
In spite of this revolutionary change in the communication process, this world remained linear, unidirectional and static, where users only surfed the Web to look for information or built fixed websites (Cormode & Krishnamurthy, 2008). However, several technological advances led to the development of a more dynamic environment at the start of the twenty-first century. New protocols (SOAP), languages (XML, RDF) and formats (RSS) were developed by the industry to facilitate the expansion of electronic commerce on the Web. This sector demanded spaces online where commercial transactions were easy, fast and safe. In this way the Web was converted into a platform for services from where users could now not only search for information, but carry out any type of action (Jarvenpaa & Todd, 1996). Now, we have changed from searching for flights to buying the ticket, from knowing the requirements for a service to directly applying for that service and from visiting a friend’s homepage to looking at the postings on their wall.
The concept of Web 2.0 emerged to describe the great changes that these new solutions were already bringing about on the Web (Knorr, 2003; O’Reilly, 2007). The concept pointed out that these alterations were not just improvements and upgrades but were going to modify the way in which users interacted with the Web and, even more, the way in which society itself was being transformed by the Web. Under this new transactional environment, the Web also started to produce new spaces where users could participate in the production of content. If users can already buy goods online or file paperwork with the government, now they can exchange and manage content with other partners as well. Wikipedia (2001) could be considered the first collaborative enterprise that attempted to create a universal encyclopaedia with entries written by anyone and on any issue in the world. Its model flawlessly represented the spirit of Web 2.0, the creation of a self-managed information system in which the contents are created by an online community of altruistic members. The success of Wikipedia, with more than 5 million entries today1 demonstrated that collaborative actions could achieve great purposes without the supervised oversight of publishers, distributors, content companies, etc. This collaborative effort was extended and new applications were derived from the same paradigm. Delicious (2003), perhaps the best example of the new Web 2.0, was created as a service to label or tag viewed content on the Web. As a kind of bookmark page, this platform allowed the creation and management of personal libraries of favourite web resources. Although functionally this had already been implemented by all the web browsers, the great contribution of this platform was the utilization of two key elements that would define the upcoming social services on the Web. Firstly, these personal collections of references could be shared with other members, thus allowing these posts to be reused by the online community, which could edit, correct and comment on these same items again. The second innovation was that those references were organized using keywords or tags that each user freely added, creating an interconnected system of categories that structured these contents. This networking behaviour generated a global knowledge system in which the information would be produced and categorized using the collaborative will of the people. In the same way, hundreds of sites emerged applying this philosophy to any type of materials. Social platforms for sharing videos (YouTube, Vimeo), photographs (Flickr, Instagram), music (Last.fm), news (Digg, Reddit, Slashdot), messages (Twitter, Tumblr) and documents (Scribd, SlideShare) appeared everywhere extending this model to any facet of life. This indexing method was not exclusive to social networking sites but was spreading to other information systems such as directories and search engines. The most interesting thing is that this model shaped a new social awareness of the importance of the community in the production of content and the power of the group to filter and select valuable information. All these platforms demonstrated that there was an important community of web users interested in collaborative projects and had evidence that social networking would produce successful and profitable products.
1.2 Social Networking Sites – The Web of the People
In this context, platforms were created whose only purpose was to put users in touch with other users. The first social networking sites, Friendster (2002) and MySpace (2003), functioned as personal directories where their members could meet other friends through the network of acquaintances. However, the first genuine web service that would change the concept of online social networks was Facebook (2004). Born as a restricted network for American scholars, its spread worldwide did not commence until the restrictions to signing up were removed in 2006. Five years later, it became the largest social platform with around 1 billion users (Ostrow, 2011). Its success fundamentally rested on the fact that their profiles were not just members’ calling cards but that they constituted a real space where users could express themselves posting texts, pictures, videos, etc. To some extent, these personal pages could be a kind of personal diary open only to a specific network of intimates that contained, in a multimedia form, all the main events in the lives of their users. This scheme, in which the content production was fundamental for establishing contacts, was disseminated and new specialized spaces, addressed to a specific public, were born. Vertical social networks (Lieb, 2013) now constitute the next challenge for online social network analysis and new specific services for small businesses (Wave), professionals (LinkedIn), programmers (GitHub), engineers (Spiceworks) and physicians (Doximity) are springing up everywhere.
But what is a social networking site (SNS)? Boyd and Ellison (2007) define ‘social network sites 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’, considering that ‘the public display of connections is a crucial component of SNSs’. However, although these elements, I think, could constitute an online social site by themselves, they are not sufficient to be a successful site. As can be seen in the above examples, it is fundamental that these social platforms incorporate instruments for producing and as well as adding content. In this sense, a social networking site could be defined as an online environment where users, besides creating personal profiles and establishing contact among themselves, they can also produce and insert content at disposal of their contacts or the entire community (Ellison & Boyd, 2013). This definition attempts to emphasise the informational aspect because the networking relationships might be just a consequence of the information flowing within the network. In other words, as more documents, images, videos, etc., are hosted on the system, more networking activity will be generated. In this way, an online space only can evolve if their members are able to produce, add and share information units among themselves. Hence content is the fuel of social networking.
1.3 Open Access – Toward a New Scientific Communication
Before the concept of Web 2.0 was born and social networking sites made an appearance, one of the most critical movements almost since the start of the Web arose in the academic community. A long time ago, the academic publishing system had fallen into a severe crisis (Panitch & Michalak, 2005). The number of academic publications did nothing but increase, while subscription costs grew at a dramatic pace, far above inflation. This increase did not correspond with any significant reduction in the production costs of printing. In addition to this situation, throughout the past century a process of amalgamation among publishers along with rapid acquisition programmes for new journals caused the concentration of the system in the hands of just a few large publishing companies (Elsevier, Springer, Wiley, etc.). Those most affected in this system were the academic libraries which looked on as more and more of their budgets were allocated to the payment of subscriptions, resulting in a clear reduction of precious scientific funds to the benefit of large private corporations. This situation came to a head in 1997 when the Association of Research Libraries developed the Scholarly Publishing and Academic Resources Coalition (SPARC) and put forth a set of demands from scholarly libraries and other organizations in order to solve this critical situation and offer alternatives that could ease fair access to the scientific literature. But the scholarly community was already aware on these problems. Three years before, Stevan Harnad (1994) had launched his well-known ‘Subversive Proposal’, where he encouraged the free and open exchange of scientific literature, depositing copies of research articles in public academic servers accessible through the FTP protocol. Harnad’s proposal was not ground-breaking by his own admission – the practice was already common in computing environments – but made explicit the existence of an alternative channel by which research results could be spread apart from the traditional publishing system. Thus the Web was able to emerge as an alternative way of avoiding the established publishing system, reducing costs, shortening publishing times and reaching wider audiences. In this case, the Web favoured the elimination of mediators as well, allowing direct communication between researchers without any limitation or fee.
Two main channels were established to make effective open access to the scholarly papers. The first channel were the electronic journals (gold open access) that duplicated the traditional model but were now without subscription and offered a reduced publication delay (Odlyzko, 1997). The Bryn Mawr Classical Review (1990), Postmodern Culture (1990) and Psycoloquy (1990) are a few examples of the first electronic journals that sprang up around the Web. However, this model was not compulsory and now most of the journals have an electronic version accessible through the major payment platforms owned by the academic publishing giants (ScienceDirect, IngentaConnect, Wiley Online, etc.). The second channel (green open access) was rather different and with a better fate. It consisted of the deposit of a full text copy of the manuscript in an open repository or digital archive before the paper was edited and published by the journal (Guédon, 2004; Harnad et al., 2004). This process avoids the slow publication times and assures the peer review of journals as well as the upcoming citation count. This protocol produced the flowering of thematic repositories such as ArXiv.org (1991) specializing in physics, the great biomedical deposit Pubmed (1997) and RePEc (1997) for the archiving of economics papers. Soon, it was common to upload pre-print copies of articles to a repository before being accepted for publication in a print journal. On the other hand, institutional repositories such as CERN Document Server, the eScholarship Repository of the University of California and HAL (Hyper Articles en Ligne) are used to express the scientific power of an institution as well as demonstrating the commitment of their organizations to the transparency and democratization of science. Thanks to the Budapest Open Access Initiative (2012), a manifesto that defines the objectives of Open Access, these deposits were becoming institutionalized and achieved policy mandates that oblige the hosting of publicly funded results in open repositories – for instead, Horizon 2020 of the European Union (European Commission, 2013), the NIH Public Access Policy or the Research Councils UK (2013).
1.4 Almetrics – The Social Impact of Science
All these changes both in the new technological developments and the new ways of disseminating research outputs, have produced the appearance of new metrics that quantify the use and impact of these publications in these networking environments. The Almetrics Manifesto (Priem, Taraborelli, Groth, & Neylon, 2010) exposed the exhaustion of the classical assessment system, in which peer review and citations are slow, subjective and imprecise mechanisms of reward. Instead, altmetrics ensure a fast and collaborative way to ‘filter’ the most relevant scientific results thanks to the instant appreciation of these materials by a vast online community that comments, posts, votes, follows and downloads these results through the social platforms. Although the manifesto’s authors cannot provide any evidence of this, they suggest that the computing of these measurements would provide an alternative to the traditional evaluation system. This document thus marked the starting gun for a broad range of studies to find the meaning of these metrics in the context of research evaluation. Thus, for example, tweets (De Winter, 2015; Eysenbach, 2011; Haustein et al., 2014a), Mendeley’s readers (Bar-Ilan et al., 2012; Li & Thelwall, 2012), ResearchGate scores (Ortega, 2015) and paper downloads (Bollen, Van de Sompel, Smith, & Luce, 2005) were compared with citations. However, the results have not revealed any substantial relationships with the current bibliometric measurements and therefore it is hard to believe that they could be an alternative to the current bibliometric evaluation. Perhaps one of the problems is that the proposed almetrics include a wide range of heterogeneous metrics (tweets, views, downloads, posts, etc.) that describe very different actions and purposes (Brown, 2014), without distinguishing usage metrics from networking ones. A further problem is that they are site-dependent, that is they are influenced by the environment in which they were created (Ortega, 2015). For instance, tweets are spread according to the number of followers a user has (Davis, 2012) and Academia.edu’s views or Researchgate’s downloads are determined by the number of users and publications in the network. Another problem is that these metrics are computed in environments external to the academic world. Tweets are dispersed in a popular network which appreciates the scientific results in a very different manner (Almetrics, 2014). One final problem is that these indicators are also time-dependent, as the more time a document is in the network, the more likely it is to be cited, shared, followed, etc. (Thelwall and Kousha, 2014). Surprisingly, in the midst of these unresolved problems and with clear evidence that these measures cannot be substitutes for the present system of evaluation, two firms, Almetric (2011) and ImpactStory (2011), the latter created by a number of the authors of the Almetrics Manifesto, emerged to provide statistics on these indicators for organizations and publishers. This uncovers a clear conflict of interest between commercial profit and scientific evidence, suggesting that there are more economic interests than scientific behind this movement (Colquhoun & Plested, 2014). In any case – and apart from the doubts that arise – the evidence of scientific studies is that these alternative metrics describe a very different effect, closer to the popularization of science or their impact on society than to research evaluation. In spite of this, this new generation of indicators is opening a window on the exploration of a new and different impact of science in environments far from the traditional publishing system. To some extent, these instruments bring to light the impact that the scientific literature exercises over scholars and professionals that are outside of the academic publishing system, a different and new world far from the classical bibliometric approach (Cronin, 2013).
1.5 Social Network Sites for Scientists
Into the changing landscape of new communication developments, revolutionary transformations and controversial manifestos, a range of platforms for the benefit of scholars was born during the period 2006–8. Social sites for scholars have gained importance for the academic community because they bring together the issues described above. They support free and open access to the scientific literature, incorporate metrics that allow the tracking, impact and usage of these materials, and extend social networking beyond meetings, conferences and workshops to a virtual environment.
1.5.1 Definition
However, a clear definition of academic social sites is difficult because there is a varied range of platforms and services oriented to different types of actions. Moreover, there is no a clear agreement on how these sites are to be named. Thus academic social sites (Ortega, 2015), academic social networking sites (Goodwin, Jeng, & He, 2014; Gruzd, 2012), academic social networks (Almousa, 2011; Ovadia, 2014), academic social networking services (Jeng, He, & Jiang, 2015; Oh & Jeng, 2011) and social media for academics (Neal, 2012) are just some of the terms used to designate these sites. Nentwich and König (2014) put emphasis on the profile as the structural element and define social networking sites as the media that make possible the ‘setting up a sophisticated personal “profile” with information about oneself, such as interests and activities, within a digital space that can usually only be reached after registration’. Calhoun (2014) used the generic term ‘social web’ to refer to ‘the web sites, tools and services that facilitate interactions, collaboration, content creation and sharing, contribution and participation on the web’. Oh and Jeng (2011) state that ‘academic social networking services’ ‘are online services (e.g. online platforms and/or software) that focus on supporting online research-oriented activities as well as building social networks for scholars’, while Bullinger, Hallerstede, Renken, Soeldner, and Möslein (2010) describe it as ‘a web-based service that allows individual researchers to (1) construct a public or semi-public profile within a bounded system, (2) articulate a list of other researchers with whom they share a connection and communicate, (3) share information with other researchers within the system and (4) collaborate with other researchers within the system.’
In our case, an approximate definition of the scholarly social site is formulated according to the essential capacities that they have to offer. Thus a social network site for scientists has to be an online space that generates statistics on its usage and the activity of its members contributing academic contents and interacting with other members. This definition considers four basic elements for building an academic social site: (1) profiles – through which a user can participate and interact in the network; (2) contents – the materials that are supplied or produced in the site; (3) networking – the connections that profiles make among them; and (4) the metrics –the measurements that quantify the actions performed in the platform. This last requisite is not indispensable but it constitutes a valuable object for the attraction of scholarly users. Hence content is at the centre of this definition because it is the instrument that articulates the relationship between the remaining elements.
1.5.2 Functions
In relation to the definitions above, several authors describe the principal functions that these platforms should develop. Codina (2009) describe three components: document management, academic profiles and groups. Bullinger et al. (2010) detect four main functions: identity and network management, communication, information management and collaboration. Oh and Jeng (2011) detail three basic functions: building a profile, management of personal publications and provision of a platform for online group research activities. Nentwich and König (2014) detail eight functions that a social site should accomplish: profiles, communication, networking, ‘directing attention’, groups, calendar, literature-related functions and further services. Many of them are easily dispensable nor are they exclusive to these vertical platforms. Espinoza-Vasquez and Caicedo-Bastidas (2015) found five actions that these sites must permit: collaboration, online persona management, research dissemination, documents management and impact measurement, and distinguished research dissemination and document management as services addressed to the contents contribution. In all these cases, these functions could be reduced to three basic types operations: a profile that identifies the user, instruments to put up and generate contents and an environment to share those outputs.
1.5.3 Motivations and Adoption
Many studies have approached the analysis of academic social sites from a qualitative point of view, exploring through surveys and questionnaires the perception of the academic community of these tools and the value they put on these sites for their research activities. A report from the Research Information Network (2010) defines two main benefits from the use of Web 2.0 services: communication with the research community and the support of colleagues in the use and adoption of new methods and techniques. Gruzd and Goertzen (2013) detected three benefits of using academic social sites: information gathering, collaboration and information dissemination. Other studies have discovered that in response researchers emphasize collaborative activities as the main benefit and utility (Jordan, 2014; Cann, Dimitriou, & Hooley, 2011). In this sense, Van Noorden (2014) revealed that most of the respondents used Academia.edu and ResearchGate for purposes of contact. Independent of these benefits, the ratio of adoption is quite low today (Procter et al., 2010) which could be cause by the absence of any immediate benefit, difficulty or reticence in the use of these platforms (Coppock and Davis, 2013). However, many authors have detected differences in adoption rates, mainly according to age (Park, 2010). In many cases, researchers adopt only one or two profiles at most in these platforms (Mas-Bleda, Thelwall, Kousha, & Aguillo, 2014; Haustein et al., 2014b; Ortega, 2015).
1.5.4 Typology
Social network sites for scientists are a heterogeneous set of applications that use different methods to promote interaction between their users. Bullinger et al. (2010) define four types of academic social network: research directory sites, research awareness sites, research management sites and research collaboration sites. Oh and Jeng (2011) just distinguish social networking sites from web-based social software. Nentwich and König (2014) distinguished different types of social network sites according to three criteria: intended usage forms, requirements for usage and available communication forms. In this study, academic social sites are grouped according the type of content and the way in which it is managed:
scholarly directories – there are just lists of user profiles (i.e. BiomedExperts, UniPHY);
social bookmarking sites – these are sites in which their users post and tag academic web resources (i.e. CiteULike, BibSonomy, Connotea);
reference management sites – these are spaces where the principal activity is to share bibliographic references (i.e. Mendeley, Zotero, Qiqqa, Papers);
document sharing sites – these platforms are addressed to share the academic outputs of their own users (i.e. ResearchGate, Academia.edu, Figshare).
1.5.5 Business Models
The building and start-up of an academic social site require an important economic effort that ensures the viability of the platform. Many of these sites started as student projects or experimental prototypes that requested funds from investors to initiate the first steps. Academia.edu, ResearchGate and Mendeley were financed by venture companies (Spark Capital and True Ventures), foundations (the Bill and Melinda Gates Foundation) and angel investors. The success of a site also has to be supported by a clear and defined business model that ensures its economic continuity (Peters, 2013). This does not mean that the model has to produce monetary benefits, but that the income must guarantee that the service will continue working with total normality. The importance of this fact is not founded on the site’s own needs but on the fact that it contains items that many users have deposited and therefore the service should ensure access to these personal materials. Different approaches are used to gain income that makes possible the working of the network. Academia.edu opts to publish announcements of academic positions and, in the near future, will be offering an advanced stats service addressed to academic institutions to discover early impacted works (Shema, 2012). ResearchGate follows a similar approach with the publication of job offers supplied by Academic Jobs. Elsewhere, CiteULike is financed by ads from the AdWords service provided by Google as well as by subscriber members (gold) who pay for enhancing the storage space and access to specific services. Mendeley ensures its funds through an agreement with Elsevier as well as developing a paywall model for premium and institutional users who can access advanced functionalities. Meanwhile, BiomedExperts and UniPHY, developed by Collexis, were ending products that were sold as a block to academic institutions. Nature Network was a product entirely developed and supported by the Nature Publishing Group. Only BibSonomy and Zotero do not have a business model but are funded by academic organizations. These different business models are an example of the newness of these services and the difficulty of developing the optimal economic model for these products. In addition, several voices have set out ethical doubts on the monetary benefit of these platforms because they utilize user-generated contents for third parties to commercialize (Arvidsson and Colleoni, 2012; Fuchs, 2010). Other ethical problems arise when many of these services are constituted as private firms (i.e. ResearchGate and Academia.edu) that encourage open access, thus taking economic advantage of a public movement.
1.6 Methods
The development of a quantitative study entails a precise and detailed description of the instruments, materials and sources used to extract and analyse the data.
1.6.1 Scope
This study is limited to a selected range of specialized social networking sites for scientists. Thus popular networking platforms such as Twitter, Facebook or Figshare, commonly used by researchers, were excluded because they do not specifically address the scholarly community. An analysis of these sites would show a distorted view of the academic activity and the results could not be narrowly representative of the scientific communication process. In addition, paywall systems such as EndNote were excluded because they limit access to non-customers. Google Scholar Citations was also excluded because, although it contains profiles, it lacks networking utilities. The two most representative sites from the previous typology were selected because of their popularity and their extensive use in the scholarly community. Two sites were selected to allow comparison. Thus BibSonomy and CiteULike were analysed as representative examples of social bookmarking sites, Mendeley and Zotero as reference managers and ResearchGate and Academia.edu as examples of document sharing sites. Nature Network and BiomedExperts are also described because they were pioneering platforms.
1.6.2 Indicators
Several indicators are proposed in this study (see below) to describe the performance of each social platform. One of the advantages of a quantitative approach is the ability to develop indicators that make possible measurement of the activity carried out by the network and thus make a fair comparison between platforms.
1.6.2.1 Activity
Activity refers to the proportion of items posted to the network by the number of users registered. Hence this indicator expresses to what extent users add content. This measurement is also calculated for groups and forums in contrast with the global activity on the platform.
1.6.2.2 Compound Annual Growth Rate
The compound annual growth rate (CAGR) is an indicator which measures the mean annual growth rate of a value across a time period. It is used to calculate the rate of increase of each social site according users, posts and publications. This measurement is most stable in exponential growths. The formula is:
where V is the value in the initial moment (t0) and in the final one (tn). Commonly, it is interpreted in percentage terms.
1.6.2.3 Country Penetration
Country penetration is ratio of the percentage of users on a site and the percentage of researchers employed in R&D by country. This measure attempts to evaluate the success or failure of an academic site in a country. The calculation allows a reduction of the size effect of large countries taking up scholarly social platforms. The information on the number of human resources appointed to scientific activities was obtained from the UNESCO Institute for Statistics (2015), being the last available data from 2011, although some countries only present data from previous years. This ratio of percentages was used because these statistics do not include all the scholarly community of a country since they exclude students and other professionals. It is thus presupposed that the proportion of those in R&D in a country could be similar to the total percentage of social network’s profiles in that same country.
In this way, penetration of a site (i) in country (c) is the proportion of users of that country (uc) in the entire site (Ui) divided by the proportion of total researchers from that country (nc) in the total amount of researchers in the world (N). A penetration beyond 1 shows that the proportion of researchers in that site is higher than the real world. Inversely, a penetration below 1 means that the site contains less researchers from a country in relation to the same proportion globally. The result is not a percentage and has to be interpreted as the number of times that one percentage is larger than another. For example, a penetration of 2.12 means that the percentage of users on a platform is 2.12 times higher than the same at world level.
1.6.2.4 Country Spreading
Country spreading (CS) is the accumulated percentage of users belonging to the first ten countries in each site. Thus a site with an elevated CS shows that the first ten countries contribute the majority of the users and, therefore, demonstrates that the site has not spread far. In contrast, a site with a low CS shows that there is a large number of users that belong to a range of countries which indicates that the site is globally spread. This metric allows us to make comparisons across platforms.
1.6.2.5 Recent Activity
This indicator tries to measure the percentage of content supplied during the period 2014–15. As more items are added to the platform in this period the more up to date and recent is the platform. This measurement allows us to observe sites that are becoming stagnant or spaces with strong energy.
1.6.3 Sources
In this study, a manual inspection of the respective social networks was undertaken to describe their functionalities and services. In particular, their blogs and help pages were retrospectively explored to find out when their functionalities were implemented or ceased. As much as possible, other sources such as bibliographies, scholarly databases and academic search engines were queried to gather the most exhaustive academic bibliography on these online services. A special mention is the use of the Archive.org WayBack Machine to find information on the evolution of these sites according users and documents. Archive.org is a non-profit organization that attempts to archive the most important pages on the Web. Thanks to the WayBack Machine, it is possible to explore successive snapshots of these pages across time, showing how these sites have evolved. On many sites, on the main page the number of registered users, number of documents, groups, etc., in that moment in time were frequently reported, so the WayBack Machine was considered a suitable tool to observe this evolution.
1.6.4 Data Extraction
A critical element in a quantitative analysis is the numerical data that describe the performance of the academic sites studied and the processes required to extract and harvest this information. Generally, several crawlers were designed to extract the desired data. Using a screen scraping technique, SQL scripts were written to navigate across the site architecture and extract the pieces of information needed for this study. 25 virtual machines from the Cybermetrics Lab were used to extract these data. In the case of BibSonomy, however, an API was used to extract information on users and posts. In other cases, such as Mendeley, these instruments offered limited possibilities and were discarded. In the case of services that have disappeared such as Nature Network and BiomedExperts several bibliographic sources were used to obtain a picture of their function and structure.