A Webometric study of selected academic libraries in eastern and southern Africa using a link analysis approach

This paper investigates the Web presence and performance of academic libraries drawn from six countries in eastern and southern Africa (i.e. three each from eastern and southern Africa). Using link and content analyses, the study evaluated the libraries’ websites in order to: examine the location of the libraries’ websites in the university website; measure each library’s link performance in terms of web page, directory, domain and site out-links and in-links; examine each library’s performance in terms of the provision of essential online services (e.g. web-catalogues, web forms, web search engines, etc.); map the libraries’ inter-linkages with each other; and to find out the most targeted web-sites/pages. Results, discussions and recommendations are provided.


Introduction
Worldwide, library practices are undergoing major revolutionary changes courtesy of the emergence of the Internet.Librarians can now perform several tasks through the Internet.Some of these applications include (a) Electronic Mail (Email), which librarians can use to communicate with colleagues and other clients and participate in discussion groups where they can share experiences and ideas with other librarians and create and monitor discussion groups of interest to their patrons; (b) Telnet, through which librarians can connect to remote computer resources and explore other library catalogues, access database services, and share resources; and (c) File Transfer Protocol (FTP) which librarians can use to obtain software programs, text images and sound files on their and their clients' behalf (McKenna, 1994).Several authors have noted the opportunities that the Internet offers librarians and libraries (Ahmed, 2002;McKenna, 1994).McKenna (1994) outlines several of these opportunities and benefits which libraries and librarians can enjoy when connected to the Internet thus: leadership opportunities (leadership in the introduction of Internet services to the industry and community, and user training); cost and time savings; question answering services; international interlibrary loans; document delivery services; online transactions; provision of government information; and information sharing.
Libraries in Sub-Saharan Africa have embraced the Internet quite recently and as such they are still struggling to measure up to the Internet-related challenges such as the construction and development of library websites, subscription to online resources (journals, databases, e-books, etc.), implementing their Online Public Access Catalogs (OPAC)sometimes referred to as Webcatalogs -online communication (with patrons, suppliers, etc.), online library acquisitions, library networking (interlibrary loans), etc.All these services require online connectivity on the part of libraries and thus library websites are increasingly becoming, most importantly, a pre-requisite for libraries to offer and market their services online.Partnerships or consortiums, too, among libraries, and between libraries and other institutions, through the Internet are deemed necessary for effective management of their services.The realization of the Internet-related benefits and opportunities, therefore, depends not only on a library's connectivity to the Internet but its Web presence, structure and impact.It thus becomes crucial to audit the libraries' websites periodically with a view to measuring their web presence, structure and impact.

Purpose
This study sought to audit and map selected university libraries' websites in order to measure the libraries' web structures, content, and visibility/presence.In view of this, the study's focus areas included the following: the number of web pages, in-links, out-links, location of library links on universities' websites, and the most popular link(s) targeted by these universities.The most popular sites were examined with a view to determining the institutions/organizations/sites with which these libraries are linked and/or partner.
It is notable from this brief review of the literature that little has been achieved in evaluating libraries using link analysis.Link analysis is a measure of the quality of and relevance of the set of links pointing to a given site (www.2020visible.com/re-terminology.html).This analysis is meant to assist in determining websites that do promote a given website and how popular one's website is.Link popularity measures how many other sites indexed in the same search engine have links to one's site by count of links to that particular site.This method has been used to measure the universities' web impact factors (Thelwall, 2002a); find the most important university web pages (Thelwall, 2003); identify link relationships between universities (Thelwall, 2002b;2002c); classify link types in academic environments on the Web (Bar-Ilan, 2005); and to rank universities (InterLab, 2005).

Methods and tools
The following procedures were used to identify academic libraries whose websites were the subject of the current study: 1.Selection of countries: At this stage, a total of six countries, out of a total of 18 countries (excluding islands) in eastern and southern Africa, were purposefully selected for study as follows: Kenya, Tanzania and Uganda (Eastern Africa) and Botswana, South Africa and Zimbabwe (Southern Africa), i.e. three each from eastern and southern Africa.2.Selection of universities: universities in each of the six countries were identified using several sources that included: A total of 59 universities which had own websites were identified from among a total of 65 universities in the six countries.3.Selection of libraries: It was intended that all academic libraries in the six selected countries be surveyed.As a result, each of the university's website was browsed for information regarding the library's home page URL address.This address was important because it was the one that was used to measure the libraries' performance on the Web.Upon browsing the universities' websites for libraries' URL addresses, only 36 libraries were selected for study based on home page ownership.Excluded from analysis were academic libraries a) whose URL home page addresses were so long and contained punctuation marks, (e.g.http:// www.unisa.ac.za/Default.asp?Cmd=ViewContent&ContentID=17) b) whose home pages were still under construction.A complete list of the academic libraries in the six countries is provided in Table 1.The last column provides notes on the reasons for exclusion of some of the libraries from the final analysis.
Relevant data were extracted from the Internet between 5 th and 20 th July 2006 by crawling the web using the SOCSCIBOT computer software (developed by Prof. Mike Thelwall).The extraction of data was conducted only once for each library in an exercise that was so slow and sometimes took as long as three days (day and night) in the case of some libraries, especially those that had large websites.Data were analyzed using the SOCSCIBOT toolkit that consists of matrix and pajek tools in order to: 1.find out the volume of each library's total web pages 2.find out the volume of each library's total out-links 3.measure each library's in-links from and out-links to other libraries in the six countries 4.measure each countries total library web pages and out-links 5.calculate the average number of web pages and out-links per library in each country 6.rank libraries according to the number of pages, out-links and out-links per page.The Pajek computer software was used to prepare the visual networks.The location of a library link on the university's home page was evaluated so as to assess the ease of accessibility of the library's website.It was noted that out of the total 36 library websites surveyed, 34 (94.4%)library links were provided on the respective university home pages.Only two (5.6%) library websites had accessibility through the academic programs' directory.
forms while majority of the libraries (30 or 83.3%) posted special library guides (orientation and training resources) on their websites.

Number of pages and out-links
Table 3 provides the rank distribution of the libraries according to the total number of web pages, out-links and out-links per page.There were a total of 22508 web pages and 275775 out-links, a situation that generated an average of 12.2523 out-links per page.Dividing the total number of pages and out-links by the total number of libraries yielded 625.2 web pages and 7660.4 out-links per library.On overall, the RU library is ranked number one with 628 pages, 19694 out-links and 31.3599out-links per page followed by [all in the order of pages, out-links and out-links per page] CPUT library (335, 9774, 29.1761), SUN library (917, 16784, 18.3032), 9578,43.9358),and the UCT libraries (16468, 173616, 10.5426).Ranking number one according to the total number of pages was the UCT library followed by UP, SUN, UFH, and RU libraries in that order while the top ranking libraries in terms of the total number of out-links were UCT, RU, SUN, UP, and CPUT.The UKZN -HOWARD libraries topped the list of the libraries with the most out-links per page (i.e.43.The total number of library web pages and out-links were computed for each country in order to find out the most productive country.South Africa was the leading with 21793 library web pages followed by Kenya (486) (31).Adobe is a company that develops a variety of computer programs such as the Adobe Engagement Platform -built around Adobe PDF and Flash technology -that greatly enhances how people create, manage, deliver and engage with information.The National Library of Medicine (NLM) on the other hand, is a medical library based on the campus of the National Institutes of Health in Bethesda, Maryland.The company collects materials and provides information and research services in all areas of biomedicine and health care.One of its well known products is the MEDLINE database.
The National Services Corporation (NISC, USA) provides information products that are accessed through BiblioLine, a Web search service, or on CD-ROM.BiblioLine is a straightforward yet sophisticated approach to online searching via the internet.SABINET is the short form for South African Bibliographic Network.SABINET facilitates electronic access to information, largely through the provision of bibliographies while ScienceDirect is a an online database that offers more than a quarter of the world's scientific, medical and technical information online.Its coverage include over 2,000 peerreviewed journals; hundreds of book series, handbooks and reference works; and back files.
An examination of the top level domains indicates that the .comor .coTLDs were the most common, recording 59 hits (i.e.79.7% of the 74 top ranking sites or web pages).In the second position was .edu(4 or 5.4%).The .gov, .netand .orgTLDs recorded 3 hits each.
It was observed that there were a variety of sites that are targeted by E&S African academic libraries.When classified according to the types of services or products these sites provide, the following categories emerged in the top ranking web sites: news sites (e.g.nytimes.com,daily-mail.co.zm, etc.), search engines (e.g.google.com),electronic databases (e.g.sabinet.co.za), database publishers (e.g.nlm.nih.gov,sciencedirect.com,etc), computer software companies (e.g.university received and produced 4.17, 2.75, 1.69, and 1.67 page, directory, domain and site out-links, respectively, as illustrated in Table 6.Another notable observation from Table 6 is that most (approximately 50%) libraries did not have any in-links from and out-links to other libraries under the scope of this study.

Libraries' networks
When plotted on a visual map, Table 4 above yields figures 1, 2 and 3.The nature of the directory links was identifiable.Some of the linked directories are concerned with library services, information about the library, information literacy, resources, digital reference library sites, electronic resources, library collections, etc.
Finally, the libraries' web pages network reveals several inter-linkages.Fig 3 and 4 provide a map of the web page linkages with and without labels, respectively.Both illustrations demonstrate dense networks that originated largely from South African academic libraries.There were a total of 8 networks belonging to libraries in South Africa while Tanzania and Zimbabwe produced one network each.

Discussions, conclusions and recommendations
Notably, libraries in Eastern and Southern Africa are well aware of the benefits and opportunities of the Internet and the Web.Despite the digital divide and technological barriers in the region, librarians in the two regions have expedited the construction of library websites.Out of a total 59 universities in the six countries that had their own websites, only 9 (15.3%) did not have library websites.Information available from three university websites indicated that the library websites were under construction.
It is our conviction that the location of a link to the library's website on the university's website/home page should be seriously considered when designing the website especially for accessibility purposes.A library whose website is hidden is as good as a library that does not have a website.It was encouraging to note that majority of the libraries' websites had links from the respective universities' main home pages and that these links were well labeled with the term 'library'.To illustrate this more clearly, take for example a library's link that is provided on or under a university's facilities' website.Library users are left to guesswork when accessing such a library's website.Not only does this situation inconvenience a user, but can lead to non-usability of a library's services, resources and products.It was observed in this study that accessibility of three libraries' websites was a bit problematic.Links to these libraries' websites were provided under the 'academics' website.One university provided one page of a description of its library services under the directory heading 'facilities', implying that the library is just as any other facility in the university.Some of the 'other' facilities that were described alongside the library were sports and recreational facilities.It is comforting, though, to note that the university does not have a library website and perhaps that is why the university chose to provide information regarding the library in the way it did.Most probably, when the university library website is constructed, measures will be taken to appropriately provide links at the right place, preferably on the university's home page.
Concerning the provision of essential web-based services, it was noted that the online public access catalogs were the highly regarded service by the libraries most probably because they are valuable tools in accessing a library's collection.Equally seen as essential library services are the guides (or orientation and training resources).Little regard is placed on search engines.Web forms were the least popular.Whereas OPACs and guides are library-specific (i.e.specific to a library offering them), search engines and web forms are not and perhaps that explains why their provision on the libraries' websites is not deemed necessary as they can be accessed through other means.For instance, search engines such as Google can be accessed through its google.comwebsite.However, libraries are advised to provide these services on their websites so as to make it easier for their clients to navigate the Internet from their site.Ahmed (2002:266) observes that "library users value the services that they access from their desktops because the services save time".Concerning the web forms, the author argues that web forms make it easier and convenient for users to communicate to libraries.Communication can be in the form of comments, suggestions, and requests for instruction classes, etc.
The ranking of libraries according to the total number of web pages, out-links and average out-links per page led to the following observations: • South African university libraries performed better than their counterparts in Botswana, Kenya, Tanzania, Uganda and Zimbabwe • No university library had a uniform rank throughout the three variables • Libraries that had branch libraries registered higher number of pages and out-links.For instance, the Howard College Campus library of the University of KwaZulu Natal comprises four libraries, namely, EG Malherbe (Main), Barrie Biermann (Architecture), GMJ Sweeny (Law), and Eleanor Bonnar (Music) while the UCT library consists of the main library and nine branch libraries.On overall, Rhodes University library was the top ranked followed by Cape Pennisula University library, Stellenbosch University library, KwaZulu Natal (Howard College) libraries, and University of Cape Town libraries.Perhaps librarians from other academic libraries in the region could be advised to learn from the Rhodes University library whose web performance supersedes others.
Similarly, the distribution of the total number of pages, out-links and average out-links per page showed that South Africa was the most prolific country followed by Kenya, Uganda, Botswana, Tanzania and Zimbabwe.Whether this pattern is attributable to technological advancement and/or the availability and implementation of information technology and communication (ICT) policies in the respective countries was not observable from the analyzed data.But recent developments in the formulation of ICT policies in Kenya and South Africa could explain the two countries' performance.There was no information that was readily available regarding the other countries' strategies in the formulation of ICT policies.
The most commonly targeted websites and TLDs include adobe.com, nlm.nih.gov, biblionline.nisc.com, thor.sabinet.co.za, sciencedirect.com, gateway.ovid.com, serialssolutions.com, search.rdsinc.com, home.ncifcrf.gov, scirus.com, etc.Among the 74 top ranked websites and TLDs were online newspapers' sites (e.g.nytimes.com-USA, daily-mail.co.zm -Zambia, sowetan.co.za -South Africa, suntimes.co.za -South Africa, sundaymail.co.zw -Zimbabwe, etc.), electronic databases and database publishers (e.g. nlm.nih.gov, biblionline.nisc.com, sabinet.co.za, sciencedirect.com, isiknowledge.com, etc.), and book-and periodical sellers (e.g.serialssolutions.com, swetsnetnavigator.nl, etc.).This perhaps shows the importance that libraries place on electronically available resources which may have cut down the costs of subscription.Library resources (e.g.journals, magazines, books, reference materials, etc) that used to be only in print format are increasingly becoming available electronically thus changing the libraries' traditional ways of providing such products to their clients.This has called for online subscriptions to journals and other periodicals that were previously accessed through print subscriptions and perhaps the high ranking of electronic databases as the most commonly targeted sites and TLDs explains the changes that libraries are undergoing and have embraced.Accessing these resources, however, requires some computer softwares that are compatible with the formats in which they are published and that may explain adobe.com'spopularity among libraries.Most full-text papers are currently published in PDF format.
Figures 1, 2, 3, and 4 provide library networks based on site, directory and page in-links and out-links respectively.It was observed that South African academic libraries do have a well developed network that included all the libraries surveyed in this study.Only two libraries each from Tanzania and Zimbabwe formed networks.Contrary to expectations, none of the libraries had links with other libraries outside their geographic regions.The few networks that were generated comprised libraries from the same country.This may imply less collaborative activities between libraries in different countries, although this observation may not be necessarily true since there is "little knowledge about models and motivations of linking" (Zeinolabedinio, Maktabifard & Osareh, 2006).However, it is well acknowledged that South African libraries do have an established inter-library cooperation and that may explain the inter-linkages that were witnessed among the libraries in the country.Some libraries were excluded from the analysis on the basis of their website addresses which contained punctuation marks or were too long to be crawled by the personal SOCSCIBOT.Librarians are advised to provide simple and short website addresses of their libraries.Keeping the website address short and clear is very important, especially for purposes of accessibility by users.Library users may want to access the library's website by use of the library's website address and if the address is too long and/or contains some characters/symbols, it may become very difficult to memorize it, thus impeding accessibility.In addition its is recommended that • Libraries provide space for Open Access repositories which may increase in-links • Libraries develop useful websites that can attract users and ensure that these websites are always functional • Libraries develop partnerships with others • Libraries develop in-house ICT policies which can be entrenched in the main library policy(ies) Finally, it is recommended that librarians become more involved in the construction of their libraries' websites.They should actively participate in policy formulation as regards legislation of internet-related laws and in the organization of services and resources on the internet.Libraries need to regularly update their websites to keep up with the current proliferation of Internet-based resources that are increasingly becoming freely available.For instance, the current agitation for the availability of journal articles through open access provides opportunities for libraries to improve on link development.Library cooperation (e.g.resource sharing, inter-library loans, etc) needs to be encouraged among academic libraries in Kenya, Tanzania, Uganda and Tanzania so as to improve their web linkages.Further studies are recommended so as: 1.To cover all academic libraries in Sub-Saharan Africa 2.To study the coverage of African academic libraries in major search engines and other online indexing services 3.To use other performance measurements such quality, relevance, etc. in evaluating library websites.4.To study ICT policies that guide in the construction and maintenance of library websites in Africa

Figure 1 Figure 2
Figure 1 Academic library networks: site in-and out-links

Table 1
Countries, universities and libraries selected for study

Table 4
Key: r = rank, f = frequency (i.e.number of pages, out-links and out-links per page, respectively) Distribution of library web pages and out-links by country .com),government sites (e.g.gov.za), and electronic books and general reference resources (e.g.yourdictionary.com). adobe

Table 5
Most commonly targeted web pages and TLDs Table6ranks the libraries according to the total number of page in-links.The Table provides the total number of in-and out-links (page, directory, domain, and site) that each universities received from or gave to the other universities, respectively.For instance, the UKZN [HOWARD] libraries received a total of 40 page links, 8 directory links, 5 domain links and 5 site links from the libraries that were surveyed, while it offered no out-links to any of the libraries.The UCT libraries on the other hand, provided links to other libraries as follows: 25 page and directory out-links, 15 domain outlinks and 14 site out-links.Several university libraries (e.g.MAKERERE, BU, ANU, CUEA, EGERTON, MASENO, UEAB, USIU, UDSM, AFRICAU, NUST, and KABARAK) did not receive links from nor provide links to any of the other libraries.Overall, there were a total of 150, 99, 61, and 60 page, directory, domain and site in-links, respectively.On average, each

Table 6
In-and out-links among the academic libraries in the six countries