Bruce D'Arcus wrote:
> Incidentally, some of the things I've complained about in the design
> of MODS and MADS (most particularly the heavy reliance on often
> optional attributes for important semantic content) are precisely
> those things that conflict with an RDF representation (though this was
> not why I was complaining).
There's another design issue to highlight. Taking Dublin Core and
PRISM  both of these specifications place an emphasis on defining
controlled vocabularies for publishing metadata rather than on defining
The PRISM group have recognised one particular use case (content
aggregation) and created a specified document type (PRISM Aggregator
DTD) that assembles that vocabularly in a well-defined XML format.
This approach has allowed the reuse of PRISM in other scenarios,
e.g. within RSS 1.0 feeds, as pioneered by the folks at Nature.
DC has similarly been reused in other contexts.
From what I've seen so far, MODS doesn't encourage this kind of
usage (please correct me if I'm wrong!)
From my perspective it seems like a Good Thing is library metadata
can be freely intermixed with data from elsewhere on the web
(e.g. from the growing number of "social content" services). RDF
provides a framework for this.