Re: Semantic content negotiation (was Re: expectations of vocabulary)

Xiaoshu,
   I was refuting your claim that "The only thing that is not  
dereferenceable is literal values". This is far from accurate. Good  
practice is quite another thing entirely.

   Both Henry and I already pointed out that the FOAF docs will  
return RDF if you ask for it.

   The hash-or-slash debate *is* relevant here. Dereferencing

     http://example.com/ontology#myProperty

   fetches

     http://example.com/ontology

   which is the base ontology URL, and looks at the fragment ID on  
the client side. Dereferencing

     http://example.com/ontology/myProperty

   tries to fetch an entirely different URL, which in my case is  
highly likely to be a 404. If you choose slashes, you either need a  
smart server (e.g., URIQA) or a smart crawler on the client. I prefer  
the former, but Apache ain't it.

-R


On  30 Jul 2006, at 7:18 AM, Xiaoshu Wang wrote:

>>> Are you sure you are talking RDF? The only thing that is not
>>> dereferencable is literal values because they are not URI.  But
>>> literal can only be an object, not subject and property.
>>
>> Not true. Only a subset of URIs are dereferenceable, and even
>> fewer are dereferenceable to yield RDF. The following are a
>> few examples:
>
> If the URI is a property and dereference it does not return a URI  
> is not a
> good practice.  I remember that the TAG is working on what is  
> supposed to be
> put in the namespace.  In FOAF's case, at least the URI is  
> dereferenable.
> But the returned type is HTML.  Here, if GRDDL is standardized, it  
> will
> still return an RDF document.  No one is required to make any URI
> dereferenable.  But the best practice should recommend so.  If an  
> RDF engine
> should follow all the links to retrieve all RDFs.


>> <dereferenceable RDF URL> + #propertyName
>> I often use slashes instead of hashes, and what you get from
>> a web server serving up a representation of that URI is
>> probably not a fragment of the ontology.
>>
>>
>
> This hash/slash debate is irrelevant here.

Received on Sunday, 30 July 2006 16:55:28 UTC