[
Lists Home |
Date Index |
Thread Index
]
Patrick Stickler wrote:
>
>..
>
> Well, I've said this before and I'll say it again here, it
> seems that the views of the Web and Semantic Web are diverging,
> or rather, the needs of the SW, particularly with regards to
> greater precision in the treament of URI semantics, is a superset
> of the needs of the Web and are thus not being met by Web-specific
> solutions.
That's not true. You don't want to accept that the Web solutions would
also work for RDF and topic maps if you just choose to use them. If you
want to make an assertion about a web page that discusses the Eiffel
tower then we agree that the two should have separate and different
addressing modes. You can either add that at the referring side, as in:
* representation:http://www.eiffeltower.com/
Of you can add that at the on the referent side, as in:
* http://www.eiffeltower.com/index.html
So you can make assertions about the *representation* all day. And you
can make separate assertions about the concept:
* http://www.eiffeltower.com/
Now you would argue that it is "incorrect" for a URI referencing the
concept to return a GET-able representation. But that is PRECISELY what
HTTP is for. It is nothing more than a protocol for GET-ing, PUT-ting
and POST-ing representations of resources and DELETEing resources.
If you want to manipulate resources directly then you need either a
totally different protocol or some kind of specialization of HTTP.
Because that isn't what it does natively.
>....
> The heart and soul of the Semantic Web are URIs, not HTTP.
> It's the ability to identify things and attach descriptions to those
> things. Some of those things might be HTTP GETable, but many,
> if not most, will *not* be (meaning all the vocabulary we use
> to organize our view of the world).
No resources are HTTP GETable. If we are talking about HTTP then I can
be quite confident in my assertion because that's what the HTTP
specification says.
Paul Prescod
|
