Describing Images (and similar), and Descriptor discovery.

Hi All,

I'm just wondering what approaches people are taking to describing non 
rdf/html resources, such as Images, PDFs and similar?

Given that we have a jpeg with the URL http://example.org/image.jpg 
would we:

give it the Identifier <http://example.org/image.jpg#this> and serve an 
RDF description via conneg

give it the Identifer <http://example.org/image#this> and again serve an 
RDF description via conneg - a possible issue introduced here is if you 
have an alternative SVG version with it's own fragments(?) (SVGTINY12[1])

give it a completely different Identifier 
<http://example.org/r/132#image> and 'link' from the descriptor to the 
image with..? (dcterms:hasFormat, sioc:link, uri:uri, link:uri, other?)

and on the reverse, how about descriptor discovery for images/PDFs etc, 
expose via the Link header (tight coupling to HTTP), or?

As a side but related question, do we see the Web Of Data as running 
autonomous to the Web of Documents, as in it is taken that rdf / linked 
data clients will purely run over the web of linked data and reference 
non rdf resources via links with no backwards discovery needed, or is 
the link from web of documents back to web of data needed by any 
specific use cases? does conneg suffice? what if the image is 
<ftp://example.org/image.jpg>?

Best,

Nathan

[1] http://www.w3.org/TR/2008/REC-SVGTiny12-20081222

Received on Wednesday, 9 June 2010 13:02:33 UTC