Browsing a Document Collection as a Hypertext

An important consequence of the diffusion of the Web is that end users are more and more getting used to browsing information according to the hypertext linking mechanism. As a consequence, not only will the number of hypertext structures increase in the near future, but also the demand of hypertext access to existing non-hypertextual sets of documents will become more and more pressing. However it is difficult to develop hypertext structures starting from an existing document collection. In this paper the authors analyze the problem of developing a link structure from a document set and show some recently proposed solutions based on artificial neural networks.