> I think there are some counter arguments to Andy Powell's comments,
> some
> of which I don't see as completely justified (some, such as length and
> complexity I see as justified). For example the fact that RDF is a
> model not a binding is well-understood by the group working on it, but
> that clearly does not show well in the documentation. What this
> proposed standard *is* doing is attempting to provide a
> semantic-web-like approach which supports Dublin Core and properties
> that are not Dublin Core whilst at the same time providing some
> backwards support for LOM where that does not conflict (a tough thing
> to
> do, politically AND technically).
Andy,
That's pretty much what we tried to do in DCMI with the DC Abstract Model - bridging the gap between the 'old' world of library approaches to cataloguing (MARC, etc - from which DC inherited much of its approach in the early days) and the 'new' world of the Semantic Web and Linked Data.
I've argued on the DC Advisory Board mailing list that the DC Abstract Model fails as a bridge because it alienates both sides. People who are used to MARC records struggle to deal with 'strings vs things' type arguments and the widespread use of URIs for stuff. People who already understand the Semantic Web just see the Abstract Model as useless fluff that gets in the way.
So I appreciate the problem.
I don't think that's any excuse for not citing the things that people are claiming to be trying to 'adopt'. As things stand currently, one can't support Dublin Core in any semantically valid sense without adopting the Dublin Core Abstract Model. This implies adopting the RDF Model. In turn that means adopting URIs as the identifiers in use. So I'd expect to see all these things being cited and used directly in the document. These are firm technical requirements.
Note that I am not commenting on the quality of this standard other than w.r.t. the question of whether it successfully supports DC and RDF. It may well be a perfectly good spec in other ways - I haven't read it hard enough to comment. I have read it hard enough to suggest (to me) that it doesn't go far enough to support DC and RDF (assuming that is what it is trying to do) other than in a very fuzzy way.
Best,
Andy.
|