W3C home > Mailing lists > Public > semantic-web@w3.org > July 2007

Re: Semantics of rdf:Statement

From: Bijan Parsia <bparsia@cs.man.ac.uk>
Date: Fri, 20 Jul 2007 19:48:17 +0100
Message-Id: <6D888930-B0DF-4635-B117-D80F37A3BA98@cs.man.ac.uk>
Cc: semantic-web@w3.org, Alexander Boer <aboer@uva.nl>, Thomas Gordon <thomas.gordon@fokus.fraunhofer.de>
To: Rinke Hoekstra <hoekstra@uva.nl>

On Jul 20, 2007, at 11:36 AM, Rinke Hoekstra wrote:

[snip]
> <rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#">
>     <rdf:Statement rdf:ID="S1">
>       <rdf:subject rdf:resource="http://www.leibnizcenter.org/ 
> information/people/rinke-hoekstra" />
>       <rdf:subject rdf:resource="http://www.leibnizcenter.org/ 
> information/people/alexander-boer" />
>       <rdf:predicate rdf:resource="http://description.org/schema/ 
> Creator"/>
>       <rdf:predicate rdf:resource="http://description.org/schema/ 
> Fish"/>
>       <rdf:object rdf:datatype=
>          "http://www.w3.org/2001/XMLSchema#string">Rinke Hoekstra
>       </rdf:object>
>       <rdf:object rdf:datatype=
>          "http://www.w3.org/2001/XMLSchema#string">Alexander Boer
>       </rdf:object>
>     </rdf:Statement>
> </rdf:RDF>
>
> The above example implicitly states

No, I don't *think* it does.

> that http://www.leibnizcenter.org/information/people/rinke-hoekstra  
> and http://www.leibnizcenter.org/information/people/alexander-boer,  
> and http://description.org/schema/Creator and http:// 
> description.org/schema/Fish pairwise point to the same resource.

This isn't an insane reading but I don't immediately see that from  
the RDF semantics. Consdier:

"""Semantic extensions MAY limit the interpretation of these so that  
a triple of the form

aaa rdf:type rdf:Statement .

is true in I just when I(aaa) is a token of an RDF triple in some RDF  
document, and the three properties, when applied to such a denoted  
triple, have the same values as the respective components of that  
triple."""

Ok, first this is a semantic extension and a MAY, so I'm not sure how  
much weight you want to give it over your reading. Second notice that  
a reification does *not* (necessarily) say anything about the graph  
in which the reification appears....it just says there is *SOME*  
graph that contains the corresponding triple (and I'm  not sure  
exactly how to interpret the existential there..some concrete graph?  
A model with a graph in it? Is every reification always true if we  
are platonists (since all graphs *exist*, including the graph with  
the single triple corresponding to the reification).

So, this would suggest that the multiple subjects might just mean  
that SOME GRAPH thinks that they are the same. (But this would not be  
expressible in RDF. You might expect a reification aware system to  
entail the requisite sameas.)

But another reading could be that rdf:subject is both functional  
*and* lexigraphicly sensitive. That is, your example would be  
*inconsistent* since there is no *syntactic item* that is a triple  
with two subjects. I think this is the more likely reading since the  
statement denotes *one* triple (per graph). And in point of fact,  
triple identity (not logical equivalence) is determined by the  
particular items in each slot.


> This is fine.
>
> However, for datatype resources this is really problematic as the  
> above RDF states that the *strings* "Rinke Hoekstra" and "Alexander  
> Boer" are equal. Something which is clearly not true...

Well that would just be inconsistent on your reading...why not?  
That's because of the built in equality theory, but you can imagine  
that you had an owl:differentFrom which made your subjects  
inconsistent too. You are proposing that the reification vocabulary  
is functional...so if it you have multiple objects which are  
differentFrom each other you SHOULD get an inconsistency.

> The W3C RDF Validator validates the above RDF as being correct. Is  
> this intended?

The RDF validator does no reasoning of any kind, afaik, so it's not  
surprising that it doesn't do an optional + non-standard extended  
reasoning :)

That is, your form is a *syntactically* correct graph, and the  
standard semantics are so weak as to always be consistent, and my  
reading of the only suggested extension in RDF semantics suggests  
both the multisubjects and multiobjects make the reification  
inconsistent. But there's clearly another reading wherein the  
subjects can be consistent and the objects not.

Hope this helps.

Cheers,
Bijan.
Received on Friday, 20 July 2007 18:48:17 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 21:45:16 GMT