The meld of keyword- and hypermedia-based information retrieval is an important research area. Keywords, extensively used to represent documents in current systems, have well-known limitations as query descriptions for users of information storage and retrieval systems. Hypermedia applications offer retrieval strategies that improve search capability but do not necessarily improve access. Models for information storage that extend retrieval access points are clearly desirable if not necessary.
Bruza and van der Weide propose a logic-based framework for a hypermedia system that incorporates structural information about the documents in the database. They describe a layered architecture that supports multiple views of the same information. They provide a calculus for information description, characterization of user queries, and depiction of the matching process. Navigation between the hyperindex and the hyperbase constitutes the retrieval process. An extended example assists understanding of their proposal.
The value of this model is its blend of navigation and representation. By extending the conceptual basis of the information store to include structural information about the document, and by embedding the match operator in the interlayer navigation process, the model produces an integrated approach to query formulation and satisfaction.
These techniques are being used in the Esprit project APPED, thus providing a testbed for practicality. Given the increasing use of electronic forms of documents as information bases, systems based on techniques such as those proposed will gain in interest.