Lists Home |
Date Index |
On Jun 3, 2004, at 9:32 AM, DuCharme, Bob (LNG-CHO) wrote:
> the semantic web is something that
> will inspire great ideas that get implemented in a different, more
For what it's worth, that's exactly the point I was trying to make in
my response to Len.
Elliotte suggests that the essential subset to do real semantic
integration will be XML+namespaces, and wonders what RDF brings to the
table. I'm a fairly hard-core RDF-skeptic myself, so I'm sympathetic.
Nevertheless, to me it is OWL that provides the mojo powering my
late-blooming interest in the semantic web stuff. Of course, OWL *is*
RDF, but it has a syntax that is more accessible and usable as XML.
For example, look at the ontologies under
http://www.w3.org/2004/02/wsa/ (the OWL formalization of the W3C Web
Services Architecture Note). Maybe I'm not the best judge, but these
seem fairly understandable to me, at least compared to other formal
notations for this kind of thing. It *was* useful to use OWL as the
formalism because we (or actually the Carnegie-Mellon people who did
the work!) could use the various OWL tools to check for undefined
terms, multiple definitions of the same term, and generally to identify
gaps and inconsistencies in the architecture.
In the day job, we have seen some real synergies between pure XML
technologies such as XQuery and semantic technologies based on OWL. I
don't have the details, but I believe we can demonstrate how a huge
ontology such as OpenCYC can be used more efficiently and effectively
by a reasoner by storing the ontology in an XML database and
selectively pulling out relevant subsets in chunks. That works best if
there is coherence between the XML syntactic structure of an ontology
and the logical structure of the network of triples. As I understand
it, that is hard to pull off in raw RDF, but more effective in OWL.
Another point I'd make vis a vis what OWL adds over raw XML is its
vastly richer notion of a relationship. Of course, XML gets a lot of
mileage with basically two built-in relationships, "has-property"
(that is, attributes on elements) and "contains" (subordinate
elements). The basic HTML hyperlink that Google exploits so well is of
course another that is easily represented in XML. I won't deny that
one can get an awful lot of benefit from these, especially when it is a
human consuming the information, but I'll suggest that OWL's much
richer relationships (and the hierarchy of support for them in
OWL-Lite, OWL-DL, etc.) prove their worth when dumb software is
consuming the information.
Again, I am only at the periphery of this stuff so I can't explain or
defend it in detail, but am just offering my impression that OWL allows
one to leverage the undeniable strengths of XML in a way that RDF (or
at least RDF-XML) does not. I would be happy to be set straight by
people who have a better grasp of all this!