On Fri, 21 Jul 2006, Xiaoshu Wang wrote:

To discribe if an IR has multiple representations is different from to
describe if it is IR or non-IR. Let's not mess this two up.  The former is
useful.  For instance, if a data is available in XHTML, WML, word, PDF, etc.
Then we can further use the content negotiation to further retrieve the
resources if using the same URI.  Or use different URIs to represent
different types/versions of representations.

Right, and it's precisely in this scenario where it would be quite useful to setup a 'mapping' from the single URI to multiple URLs, but this should be a second option to using content-negotiation against a single URL.

But it will be useless to
classify IR vs. non-IR. Of course, you can develop ontology and use it to
describe it, it just won't offer much help.

So my proposal suggests a class that defines ways of transforming
the URI you find in a SW document into URLs that get specific types of
information.

I would also be cautious about that.  This seems to be similar to what the
web service is doing.  I hope we don't try to reinvent the wheel, especially
it isn't a small wheel to invent by any means.

I can understand the caution and the same red flag was raised in my mind with this suggestion, but I guess it would help to ask if there are any specific situations where mime-types are not sufficient enough to describe 'unambigously' the representation modalities for a single URL. It seems this is the exact purpose they were set out to serve. The only situation I could think of is in trying to resolve the defining ontology of a URI, but if you first attempt to dereference it as RDF content (with application/rdf+xml,text/rdf+n3 in the accept header), the resulting set of triples should include a rdfs:isDefinedBy statement associating the URI
with it's defining ontology (if such an ontology exists).

Chimezie Ogbuji
Lead Systems Analyst
Thoracic and Cardiovascular Surgery
Cleveland Clinic Foundation
9500 Euclid Avenue/ W26
Cleveland, Ohio 44195
Office: (216)444-8593
[EMAIL PROTECTED]

Reply via email to