Lists Home |
Date Index |
Eric van der Vlist wrote:
(I am basically agreeing with Eric, and chose to reply to this message in
order to echo it and add annotations)
> Patrick Stickler wrote:
> > What we need is a global metadata registry (or registries)
> > that provides access to knowledge by arbitrary URI, which
> > is expressed in standardized ontologies with consistent
> > semantics -- ideally expressed in RDF.
> > You know, something like a Semantic Web... ;-)
> Yes, we badly need it.
I have no idea what such a "global metadata registry" is or whether this
itself would be useful. My idea of the "Semantic Web" is as an extension of
the current Web, in which HTTP (and hence REST) continues to play a big
role. Hopefully standardized ontologies will occur, but in the meantime
individuals should be able to publish their own ontologies. Consistent
semantics would be a great thing. What that means is _simply_ that machines
would be able to navigate the Web with the same ease as people. "Agents"
would be machine programs which have been instructed to behave on behalf of
a person: e.g.
"go purchase the cheapest 120 Gb drive which means my usual criteria"
"schedule me a doctor's appointment for my annual physical exam"
This sort of thing is beneficial to us, because it would increase our
efficiency and productivity.
> In the meantime, I have customers who need to publish "namespaces" or
> more precisely resources describing namespaces, you know, those nasty
> things such as text descriptions and documentations, XML Schemas (W3C or
> not), stylesheets and this king of stuff.
One vision of the so-called "Semantic Web" seems to be to entirely replace
the current Web with something else.
Good Luck, and more power to you. Call the rest of us when it is working.
When I use the term "Semantic Web" I mean an interative improvement on the
current Web infrastructure.
I will respond to this statement from Patrick Stickler's prior email:
The problem here is that if I dereference some URI expecting to
access that actual resource, and get some metadata or RDDL document
or something else in its place, how do I necessarily know that
that is *not* in fact the resource? It may not actually be apparent
from the mnemmonic qualities of the URI, and such qualities are
unlikely to be meaningful to some software application.
Let me try to say this as clearly as possible:
When you dereference a URI you NEVER get back the _resource_. Never ever.
You get back a representation of the resource. Certainly in the special case
of the "data:" scheme, the representation and the resource are essentially
defined to be one and the same, but in the case of the "http" scheme for
example, this is not the case.
The RDDL document which describes a namespace IS NOT EVER the namespace. It
is a description which MAY be returned by dereferencing the URI (e.g. HTTP
GET). This seems to follow the rules of REST quite nicely.
> I can ask them to wait for the Semantic Web, but they will probably be
> right to turn to my competitors and it's probably wise to publish their
> documents in as many schema repositories as we can find in the meantime
> (just like it's wise to submit your web pages to search engines and
> repositories untill the Semantic Web can groke them)...
> What harm does it do if they also use the namespace URI to host a RDDL
The real question is: until Patrick's all knowing global metadata registry
gets implemented and working, in the absense of something returned by an
HTTP GET, how is one to find out anything about an XML namespace? In any
case this argument is a dead dog. RDDL is designed to follow the 'rules' of
the current Web, resources and URIs.