Re: [ALL] RDF/A Primer Version

Gary Ng wrote:
> There must be situations where the href="..." is only logical and is not 
> deferenceable, or in another situations where the author really intend to 
 > just assert that resource reference as a piece of metadata and has no
 > intension for anyone to dereference?
> 
> For example,
> 
> A. Don't want to dereference for metadata: 
>    foaf:mbox     = "mailto:jo.lambda@example.org"
> 
> B. Those to dereference for more metadata (e.g in the Department members page) 
 >    to build up a contact list:
>    <a rel="foaf:member" href="http://jo-lamda.blogspot.com/">Jo Lamda</a>
> 
> C. Those to dereference but there is going to be nothing there so we need to 
> add metadata locally. In this case it seems like the approach is to use
 > additional elements with about="..." to specify.
> 

My understanding is that when loading an RDF graph from a document, 
however it is serialized, there is always the issue of whether you want 
to dereference the resources referred to, in order to get more triples.
Common practice is not to; but cwm, for instance, supports a mode where 
you do read everything, recursively ...

It is clear that if the scheme does not support a GET operation then you 
cannot, so A. unambiguously cannot be dereferenced. For case B, one 
could do a GET, with appropriate accept-headers, and if what comes back 
is RDF (in any form, RDF/XML, N3, RDF/A ...) then you could read it as 
more metadata. There is no consensus about the status of the 
relationship between the propositional attitudes expressed in the two 
documents loaded.

So I don't see this as an RDF/A issue, but a much wider semantic web issue.

Jeremy

Received on Wednesday, 25 January 2006 12:54:55 UTC