Re: need to determine what RDF is

From: Graham Klyne <GK-lists@ninebynine.org>
Subject: Re: need to determine what RDF is
Date: Thu, 30 May 2002 18:56:48 +0100

> At 11:26 AM 5/30/02 -0400, Peter F. Patel-Schneider wrote:
> 
> >From: Dan Brickley <danbri@w3.org>
> > > RDFCore 1.0 RDF/XML documents have propositional content; they represent
> > > 'claims about the world'. What factors fix this content, ie. what
> > > is relevant when we set about considering whether some bundle of these
> > > RDF/XML-represented propositions are true or not?
> >
> >Hmm, this is not really what I was getting at, as it includes lots of
> >issues about the relationship between RDF graphs and the real world.
> >
> >I'm only interested in relationships between RDF graphs.  Which such
> >relationships are RDF relationships?
> >
> >My view is that the only such relationships are RDF entailment and RDFS
> >entailment.  Any agent that computes any other relationship between RDF
> >graphs is not doing RDF.
> 
> It sure seems there is confusion about what it means to "do RDF".  I agree 
> with you that many agent functions that have been mooted go beyond what is 
> licensed by RDF-entailment.  It's difficult to imagine RDF being useful if 
> RDF-entailment is all that agents can do.
> 
> I think the importance of RDF entailment is not so much in what agents do 
> in the privacy of their stores, but in what they exchange between each 
> other.  If agent A passes to agent B an RDF graph, then agent B may use 
> only RDF-entailments based on that graph as premises to its own reasoning 
> if those premises are to be presented as authorized by A.  This does not, 
> of itself, prevent agent B from issuing further statements that have been 
> deduced using rules not sanctioned by RDF, and I agree that it would be 
> fair to say that in deducing such statements it is not doing sound RDF 
> reasoning.

I agree entirely.  However, others appear to want to call this reasoning
RDF reasoning in some fashion.

> But that doesn't seem to mean that they cannot be part of a useful network 
> of applications that use RDF to exchange information.  So it seems that we 
> do need to be clearer about what it means to "do" RDF.  For me, it is that 
> applications use RDF to exchange information in a fashion that has some 
> useful, widely recognized properties that go beyond just syntax.

Well, then someone has to define what is and, more importantly, what is
not allowable, and how to label what is going on.  I'm quite happy to say
that RDF reasoning is RDF(S) entailment, that anything else is not RDF
reasoning, and that anyone, like, for example, WebOnt, can do their own
thing, and even relate it to RDF reasoning in whatever manner they want.

> So, coming round to things like rdfs:comment - yes, I agree it's important 
> to be careful not to claim that the object of this property affects the RDF 
> entailments than can be supported by a graph, but that doesn't mean there 
> are not other ways, outside RDF, in which information can be extracted from 
> the object value.

Sure.  However, again, some people want to label this extra processing
somehow as RDF. 

> #g

peter

Received on Thursday, 30 May 2002 13:59:45 UTC