On the Destination of Retrieval Links in Hypertext Systems

Abstract

Information browsing in hypertext systems is provided by links with source and destination anchors within hypertext documents. From the point of view of link following operations, links can be static or dynamic. While for static links, source and destination are well defined before the link is used, dynamic links have no defined source and/or destination until the links are generated.

In this paper, we focus on the so-called retrieval links which use information retrieval technology to enable dynamic connections between hypertext nodes. We address the important problem of automatic determination of optimal destinations for the retrieval links within hypertext documents. A new approach based on the density distribution of important term weights within destination documents is described. The power of such a similarity measure is in local text comparison where we do not dispose of the real term weights. The implementation algorithm and characteristic examples are also presented.