Dear WG Members,
The DCMI Usage Board has recently approved a new term: provenance. This
term will be added to the Registry, DCMI documentation and RDF schemas this
coming Monday (2004-09-20).
As many of you may already know, the Registry uses reification to provide
access to provenance information. Reification, and how best to implement
it, has been given a lot of attention in the RDF community (see:
http://lists.w3.org/Archives/Public/www-rdf-interest/2004Aug/0170.html), and
for us, it is a less-than-perfect solution.
In preperation of our upcoming WG meeting in Shanghai I would like to open
discussion regarding the possible use of this term by the Registry to
provide access to provenance information. One alternative to maintaining
the provenance information locally within the Registry is to (internally)
use the provenance property to link to the source document from which this
information was originally derived. For example, the provenance information
for the French translation of term=dc:title, property=rdfs:label is:
Description Le Dublin Core est un ensemble simple d'éléments de
metadonnées pour faciliter la recherche de ressources digitales. Ce document
présente la nouvelle définition (version 1.1), en français, de ces éléments.
(Dublin Core is a simple set of metadata elements whose purpose is to
facilitate the search for digital resources. This document represents the
French translation of version 1.1).
Type (de RDF) Statement
Titre Le Dublin Core Version 1.1
Créateur Anne-Marie.Vercoustre
...
This information is displayed by clicking on the [fr-FR] RFC-3066 code that
is displayed next to the French translation of the label.
It seems to me that, rather than maintaining this information locally, a
better approach is to simply let this link resolve to the original source
document: http://www-rocq.inria.fr/~vercoust/METADATA/DC-fr.1.1.html. This
would greatly reduce the number of RDF statements that are generated for
terms, simplify the Registry processing and distribute support for the
provenance information back to it's source.
I am sure there are other approaches as well and would welcome suggestions,
comments, concerns, etc. regarding this, and any other approach, to manging
provenance information.
Best Regards,
Harry Wagner
http://oclc.org/research/staff/wagner.htm
|