> Okay, but "any other imaginable digital form" could be read to imply
> that a
> JPEG photograph of some metadata (to take an extreme example) might
> eventually
> be trivially transformable.
If the JPEG is a photo of a MARC/XML record, then the transform from a
image/jpeg into application/marcxml+xml should be trivial. Just write
the output of an OCR application to a file with a .mrcx extension.
If you were hoping for N-Triples instead, MARC needs to confront this
paraphrase of Voltaire: "If you wish to converse with me, define your
terms." RDFS/OWL vocabularies are an excellent way for domain experts to
define their terms. Unfortunately, the concepts in MARC don't translate
easily to discrete ontological terms.
In hindsight, the transition to this new ontological worldview would be
SO much easier if we had updated our 1960s era "record model" to the
1970s era "relational model" when we had the chance. Unfortunately, a
basic description of the sum of human knowledge (which libraries were
responsible for until recently) wasn't (and still isn't) scalable to a
relational model. Fortunately, the Web changes that.
> You do point out that the transformation
> could
> very well be lossy. But given the considerable intellectual and
> interpretive
> work required, can we really expect the transformation of, say, native
> MARC
> records or FRBR-based descriptions to doable so easily in the near
> term?
Yes, but part of the deal is that we have to compromise with diverse and
dynamic realities. OWL should help us do that.
> Nitpicking, perhaps: I would say that "any set of XML elements could
be
> mapped
> to RDF vocabulary". Does that capture what you mean? An apple cannot
> be
> "upgraded" to an orange...
IF an XML namespace doesn't currently resolve to anything (most don't)
and IF the people in control of the namespace's domain understand
RDFS/OWL, then it SHOULD be possible to upgrade/deprecate XML elements
to RDFS/OWL. A <dct:creator> is still a <dct:creatory> regardless of
whether the XML Schema camp realizes it or not. I would rather forgive
people for their primitive belief in XML Schema than admit that
"mapping" is necessary. For example, I wish that MADS/RDF had done this.
It's SO CLOSE. RDF is going to win eventually and besides the XML Schema
community deserves to be marginalized for inventing countless syntaxes
in the first place. :-)
Jeff
|