Classifying Web sites and Web pages: the use of metrics and URL characteristics as markers

Abstract

Points to the way in which computer scientists and librarians working with the World Wide Web are turning to traditional library and information science techniques, such as cataloguing and classification, to bring order to the chaos of the Web. Explores cataloguing opportunities offered by the ephemeral nature of materials on the Web and examines several of the latter’s unique characteristics. Suggests the coupling of automated filtering and measuring to the Web record cataloguing process, with particular reference to the ephemeral nature of Web documents and the ability to measure Uniform Resource Locator (URL) and Web document characteristics and migrate them to catalogue records using automated procedures. Reports results of an ongoing longitudinal study of 361 randomly selected Web pages and their Web sites, the data being collected weekly using the Flashsite 1.01 software package. Four basic approaches to ordering information on the Web were studied: postcoordinate keyword and full-text indexes; application of both precoordinate and postcoordinate filters or identifiers to the native document by either authors or indexers; use of thesauri and other classification schemes; and bibliometric techniques employing mapping of hypertext links and other citation systems. Concludes that off-the-shelf technology exists that allows the monitoring of Web sites and Web pages to ‘measure’ Web page and Web site characteristics, to process quantified changes, and to write those changes to bibliographic records. Capturing semantic or meaningful change is more complex, but these can be approximated using existing software.