National and international university departmental web site interlinking: a webometric analysis

Abstract

In recent years, the structural similarity between hyperlinks and citations has encouraged information scientists to apply bibliometric techniques to the Web, with the hypothesis that studies of links may reflect patterns of informal scholarly communication in the way that citations can be used to illustrate formal scholarly communication. University web site interlinking has consequently been extensively investigated, but much less is known about departmental interlinking, i. e. links to, from, or between sets of departmental web sites. University web sites are large compared with departmental web sites, and statistically significant results are more easily obtained. Nevertheless, universities are multidisciplinary by nature and various disciplines may employ the Web differently, thus patterns identified at the university level may hide subject differences. Departments are typically subject orientated, and departmental interlinking may therefore illustrate interesting disciplinary linking patterns, perhaps relating to informal scholarly communication. Similarly, international academic interlinking at the departmental level is another relatively neglected research area. The research aim of this thesis is: firstly to validate departmental link data; secondly to identify whether and how link patterns differ along country and disciplinary lines amongst similar disciplines and similar countries. In order to do so, physics, chemistry and biology departments in Australia, Canada and UK were chosen. The subjects are all hard sciences, and are therefore relatively similar, and potentially able to reveal subtle differences in linking patterns. The three countries are all economically advanced, and all are predominantly English speaking, except for Quebec, which is a French speaking zone of Canada. Techniques originally designed mainly for university link analysis are applied in this study, although modified to cope with the additional difficulties of an international department-based investigation. Both the commercial search engine AltaVista and the personal web crawler SocSciBot are used to collect the necessary link data. Significant correlations between inlinks and research quantitative indicators are present, and also between Web Impact Factors (with academic staff members as denominators) and research averages. The statistically significant results together with the results of a target page classification exercise serve to support the validity of the departmental link analysis as reflecting academic activities. Citation counts are the most relevant data to compare with link counts, since the similarity between the two triggered webometric studies. Citation counts for Australian departments from 1998 to 2002 are from Research Evaluation and Policy Project (REPP), in the Australian National University. With regard to citation counts for UK and Canadian departments, the thesis introduces a technique to count citations in a semi-automatic way. In addition to citation counts, the results from the Research Assessment Exercise (RAE) 2001 are employed for the UK departments' correlation tests, while research grants received in 2003 from the Canadian National Sciences and Engineering Research Council (NSERC) are used for Canadian departments. In order to get a holistic picture of departments' web use profiles and link patterns, four different aspects are identified for each set of departments. The four aspects are: 9 General web use " National peer interlinking " International peer interlinking " Interactions with different top level domains Different link patterns are identified along both national and disciplinary lines. Along national lines, a likely explanation for the difference is that countries with better research performances make more general use of the Web; and, with respect to international peer interlinking, countries that share more scholarly communication tend to interlink more with each other. Along disciplinary lines, it seems that departments from disciplines which are more willing to distribute their research outputs tend to make more general use of the Web, and also interlink more with their national and international peers. In summary, the country and disciplinary link patterns identified are both influenced by offline factors that affect the relationship between the source and target page owners. There can be significant differences in the way that similar disciplines use the Web, and these can point to underlying differences in informal scholarly communication.