One of the great values of RDF is that it allows for mixing and matching
of metadata elements. My syntax is probably suspect but just to slightly
tweak Alex's example:
<rdf:RDF xmlns:rdf='http://www.w3.org/1999/02/22-rdf-syntax-ns#'
xmlns:dc='http://purl.org/dc/elements/1.1/'
xmlns:ind='http://indexesrus.org/'
xmlns:rev='http://reviews.org/'
xmlns:me='http://elements.justforme.org/'
>
<rdf:Description>
<dc:subject
rdf:resource='http://purl.org/gem/instance/subject/GEM-S/arts_photography'/>
<ind:recommended_stemmer
rdf:resource='http://www.tartarus.org/~martin/PorterStemmer/'>
<ind:full_text_avail
rdf:resource='http://whereever.theheckit.is/'>
<rev:rating>
3 stars
</rev:rating>
<me:level>
toast
</me:level>
</rdf:Description>
</rdf:RDF>
You can probably extend any metadata format to cover anything if you want,
but I think that RDF has the right idea, use namespaces as a means to draw
on the work that has already been done in resource description by the
communities that might best understand the content. If I wanted to add an
indication of the recommended stemming algorithm for indexing the
resource, I am not sure it belongs in MARC, but this is arguably an
important concern if the metadata seeks to wire in the object for wider
availability. RDF plumbing also allows for mapping between metadata sets,
for example, using "equivalentTo" for equating "PreparerName" from Real
Estate Metadata with "Creator" in Dublin Core. There probably isn't that
many collections that need to work in real estate agreements, but there's
lots of scientific datasets and other non-bibliographic materials which
could probably benefit from this kind of mapping.
Although I am a total fan of mapping MARC to RDF, to me the big advantage
of RDF or Topic Maps or whatever are that they open the door to more
flexible combinations of metadata and mappings rather than that they can
render the same information in a different syntax.
art
Received on Wed May 23 2007 - 06:55:26 EDT