Results of experimentál research on automatic indexing and classification

Abstract

To provide a flexible access to digital documents in a structured form on the WWW is a basic demand on the part of users. In accordance with this prob-lem librarians alsó need accomplish a major task where they can share the accumulated knowledge with that of system developers and initiate a closer cooperation with them. The author reviews inter-nationat research projects which provide a possible solution to automatic indexing and classification of HTML documents. A detailed description is given of the Nordic WAISAA/WW project, DESIRE II project - Engineering Electronic Library System (EELS) service, GERHARD and SCORPiON projects.
The generál purpose of the projects is to explore different methods of automatic classification in order to develop resource discovery on the Web. In these projects several approaches have been applied: heuristics, weighing schemes, clus-tering and computer-linguistic methods to improve the classification process. A fruitful cooperation among researchers comes to exist which helps to refine their research methods. The established classification systems are mostly used as browsing structures in Internet services, and they rarely provide access points for specific searching. In the near future they need be changed and adapted to become a suitable browsing structure for Internet services.