The calculation of web impact factors

Abstract

This case study reports the investigations into the feasibility and reliability of calculating impact factors for web sites, called Web Impact Factors (Web-IF). The study analyses a selection of seven small and medium scale national and four large web domains as well as six institutional web sites over a series of snapshots taken of the web during a month. The data isolation and calculation methods are described and the tests discussed. The results thus far demonstrate that Web-IFs are calculable with high confidence for national and sector domains whilst institutional Web-IFs should be approached with caution. The data isolation method makes use of sets of inverted but logically identical Boolean set operations and their mean values in order to generate the impact factors associated with internal-(self-) link web pages and external-link web pages. Their logical sum is assumed to constitute the workable frequency of web pages linking up to the web location in question. The logical operations are necessary to overcome the variations in retrieval outcome produced by the AltaVista search engine.

Extracting macroscopic information from Web links

“…Almind and Ingwersen (1998) used a range of calculations to discover that Denmark was relatively less visible on the Web than other Nordic countries. Ingwersen's (1998) article that developed Web Impact Factors confirmed the previous joint findings with Almind. Snyder and Rosenbaum (1999) calculated link page totals between the global Internet top-level domains, showing the unreliability of search engine results.…”

“…A similar Web-based phenomenon has also attracted much attention in the past 5 years, that of pages containing a link to the one studied, sometimes called "backlinks" or "sitations." It has been pointed out that Web data is inherently more unreliable and technically problematical than citations (Bar-Ilan, 2001;Björneborn & Ingwersen, 2001;Cronin, 2001a;Davenport, & Cronin, 2000;Egghe, 2000;Ingwersen, 1998;Smith, 1999;Snyder & Rosenbaum, 1999;Thelwall, 2000Thelwall, , 2001a. Such analyses have, nevertheless, been conducted on e-journals and, indeed, initiatives to add hyperlinks to references in articles originally published in print may make this much more widespread (Harnad & Carr, 2000).…”

“…WIFs have normally been calculated using the advanced facilities of a search engine, a limiting factor because these cover only a fraction of the Web, and one that is biased by the very page links that WIFs use (Lawrence & Giles, 1999). Other studies have also found uneven coverage of the search database and variability of the results over time (Bar-Ilan, 1999;Ingwersen, 1998;Rousseau, 1999;Snyder & Rosenbaum, 1999;Thelwall, 2000), although an improvement in the reliability of AltaVista's results over time has recently been identified (Thelwall, 2001b). To circumvent this issue, further research used a specially constructed academic Web crawler to ensure comprehensive coverage of five universities, with WIFs subsequently calculated (Thelwall, 2001a).…”

“…An identical calculation can be evaluated for electronic journals, but also a similar one for any defined set of areas of the Web. Ingwersen (1998) proposed a link-based calculation, the Web Impact Factor (WIF). He defined the external WIF to be the total number of pages external to an area of the Web with links pointing into it, divided by the total number of pages in the chosen area.…”

Methods for reporting on the targets of links from national systems of university Web sites

“…In the context of university web sites patterns of informal scholarly communication may perhaps be traceable in a way that has not previously been possible (e.g. Cronin et al, 1998;Ingwersen, 1998;Goodrum et al, 2001) in order to complement existing bibliometric techniques (e.g. Glänzel, 2001).…”