by Thomas Baker
Some thirty experts in databases and librarianship met on 7-8 October 1996 in Bad Honnef, Germany, to discuss how to integrate access to thousands of Web resources without assuming that they sacrifice their customized metadata, format preferences, or institutional autonomy. The goal was to bring together diverse communities oriented to 'interoperability' (database experts) and 'metadata' (computer and library scientists) in order to identify objectives held in common, though couched perhaps in different terms.
The Dublin Core a simple but extensible set of descriptive metadata elements - was discussed both by two of its designers (Stuart Weibel, Online Computer Library Center; John Kunze, University of California San Francisco) and by several German organizations currently using it to catalog Web materials (Judith Pluemer, German Mathematical Society; Joachim Luegger, Konrad Zuse Center; Hans Becker, University of Goettigen Library; Diann Rusch-Feja, Max Planck Institute). What makes Dublin Core so interesting a challenge is that it needs to be both readable by humans and parsable by machines (Clifford Lynch, University of California).
Researchers described how thesauri (Michiel Hazewinkel, CWI) or ontologies (Gene Alloway, University of Michigan) or attribute models (Andreas Paepcke, Stanford University) or inheritance hierarchies of attributes (Norbert Fuhr, University of Dortmund) could be used to generate asset models (Olle Olsson, SICS) or conspectuses (Dan Kiskis, University of Michigan) for use by intelligent agents, in middleware (Markus Tresch, SGFI), or by mediators (Anthony Tomasic, INRIA). Of particular interest was the Warwick Framework for managing multiple, diverse metadata sets as packages within containers (Carl Lagoze, Cornell University).
Other speakers related these issues to catalog harmonization among European national libraries (Sonia Zillhardt, Bibliotheque Nationale de France), metadata for hypertext links (Reginald Ferber, GMD), Z39.50 (Jose Borbinha, INESC), the integration of text and relational databases (Judy Lay, RAL), and payment schemes (Jakka Sairamesh, FORTH).
The event was organized as the second workshop of ERCIM's DELOS Working Group (see ERCIM News 25). The workshop's Web page includes a list of attendees, position papers, and links to related articles. Please see under http://www.area.pi.cnr.it/ErcimDL/.
Thomas Baker - GMD