At 09:40 AM 2/21/2002 -0500, Rhyno Art wrote:
>So what does this do? Well, consider the dozens of variations of the term
>"author", and then factor in works that sort of have the concept of
>"author" but not quite, such as someone who has modified a GIS dataset.
>RDF-savvy databases can possibly let you deal with some of the many nasty
>semantic issues associated with information retrieval and, if resources
>like WordNet and MetaNet can be tapped into, allow you to take advantage of
>worldwide efforts to deal with language and communication issues.

If this were true, I could see a point to RDF, but I'll be honest; I don't 
RDF can deliver this.  RDF is a sufficiently flexible mechanism for describing
things (it *has* to be) that given any reasonably complex set of metadata, I'm
betting 5 librarians could sit down and come up with 5 utterly different RDF
encodings of the same set.  There are ways to deal with that problem.   We
come to agreement on element sets; we come to agreement on element
relationships; we come to agreement on descriptive practices.  You can then
do an RDF implementation of all that, sure.  But note that the basis of that
semantic interoperability is *the agreements*, not RDF.  And I've yet to
see a DL metadata project where those agreements existed where we couldn't
do an implementation without using RDF by specifying a DTD or an XML schema,
and have something that was easier for other libraries to adopt and use.

I think the clearest expression I can make of the belief that Roy and I 
share is that RDF isn't *unworkable*, it's just that it's unnecessary.  So, 
yes, there
are many projects out there using RDF, but are there any of them that *had* to
use RDF to accomplish what they wanted to do, or who realized benefits from
using RDF that they couldn't have achieved without it?

I suspect that the whole RDF argument boils down to another instance of the
debates about artificial intelligence; if you believe in the ability to 
create real
semantic/pragmatic understanding in machines, you believe in RDF's promise.
I feely admit that I'm in the opposite camp that believes that the AI people
need to go re-read Saussure a few times, particularly that bit about
the arbitrariness of the relationship between the signifier and the signified,
and then follow it up with Donna Haraway's work on situated knowledges.
As I said earlier, when pushed, every discussion I've had regarding RDF's 
comes back to building the semantic web; if you really don't have any faith
in that coming to pass, I don't think you're likely to see a big benefit to 

Jerome McDonough
Digital Library Development Team Leader
Elmer Bobst Library, New York University
70 Washington Square South, 8th Floor
New York, NY 10012
jerome.mcdonough at
(212) 998-2425