W3C home > Mailing lists > Public > public-rdf-dawg-comments@w3.org > May 2011

Re: (Batch) Response to SPARQL WD comments

From: Chime Ogbuji <chimezie@gmail.com>
Date: Tue, 24 May 2011 11:29:31 -0400
To: Kjetil Kjernsmo <kjetil@kjernsmo.net>
Cc: public-rdf-dawg-comments@w3.org
Message-ID: <9D53A0984E904173B0CCACCC926751F6@gmail.com>
Kjetil, See my response inline. This is not an official WG response, but just attempt to better understand your concerns with perhaps the result that a WG response is better informed by these concerns.
On Wednesday, May 18, 2011 at 11:39 AM, Kjetil Kjernsmo wrote: 
> Tirsdag 17 mai 2011 15:44, skrev Chime Ogbuji:
> > Please see the response above regarding changes made to improve the
> > readability for developers.OK, thanks! It looks better indeed, I'll make individual comments if there is 
> something.
> It is the other usage that is more important. It is problematic because it is 
> inherited from SPARQL 1.0: 
> http://www.w3.org/TR/rdf-sparql-query/#namedGraphs
> so I don't know how inclined the present WG is to do anything about it.
To be perfectly honest, the WG so far (AFAICT) is not inclined to make any changes to SPARQL 1.0 on account of this particular discussion.
> In 1.0, you could get around any problems that may have been caused by this by 
> redirecting, but now that a 200 response is required, you cannot necessarily 
> do that.
> 
> So, the problem is that according to my understanding, we are bound to get URI 
> collisions. Lets take a concrete example, my FOAF profile can be GET by 
> dereferencing http://www.kjetil.kjernsmo.net/foaf
> There, you learn that 
> <http://www.kjetil.kjernsmo.net/foaf> a foaf:PersonalProfileDocument .
> but now, it turns out that the same URI identifies some RDF graph content, 
> especially if someone says 
> SELECT * FROM <http://www.kjetil.kjernsmo.net/foaf> WHERE { ?s ?p ?o }
Note this is different from:

SELECT * { GRAPH <http://www.kjetil.kjernsmo.net/foaf> { ?s ?p ?o } } 

I.e., it is not necessarily the case that the underlying graph store will have a graph whose URI is 

http://www.kjetil.kjernsmo.net/foaf

which is really where there may be a URI collision in the way you are describing. If some RDF graph content includes some statement(s) to the effect that the graph content itself is a kind of thing that cannot be "content on the web", then it is the same ontological / web architecture paradox that you have here:

http://www.ihmc.us/users/phayes/PatHayesAbout.html

Which you can have with or without this protocol specification. 

A 200 response is only required (by the Graph Store protocol) if such a named graph is in the graph store, and (AFAICT) you can use FROM < ... > to get your FOAF profile even without this being the case, since all SPARQL 1.0 says is that the dataset subject to the query is obtained from representations for the URI and, though that URI doesn't identify graph content in the underlying graph store, some other kind of HTTP service can still meaningfully serve an RDF document in response to a GET to that identifier.
> To me (and I may be wrong, and I would love it if anybody could explain this 
> in clear terms if I am) a foaf:PersonalProfileDocument and some RDF Graph 
> Content are two distinct things.
The characteristics of a FOAF personal profile document "can be conveyed as [an RDF document that is the content of an HTTP message and] serializes a named graph paired with the graph IRI in the underlying Graph Store", so I'm not sure why - intuitively - they are necessarily two things. Can you describe why you think of them as two distinct things? Maybe you are distinguishing between information and information content? The FOAF specification doesn't make this distinction: 

"We do not (currently) distinguish precisely between physical and electronic documents, or between copies of a work and the abstraction those copies embody."

-- Chime 
Received on Tuesday, 24 May 2011 15:30:04 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 24 May 2011 15:30:05 GMT