Re: Can RDF say anything about anything?

From: Frank Manola <fmanola@mitre.org>
Subject: Re: Can RDF say anything about anything?
Date: Mon, 17 Feb 2003 10:27:52 -0500

> Peter F. Patel-Schneider wrote:
> 
> > From: Frank Manola <fmanola@mitre.org>
> > Subject: Re: Can RDF say anything about anything?
> > Date: Thu, 13 Feb 2003 14:38:07 -0500
> > 
> > 
> >>Peter--
> >>
> >>Thanks for this comment.  Regarding the situation you bring up, the
> >>Primer at that point says:
> >>
> >>
> >>>These examples also illustrate one of the basic architectural principles 
> >>>of the Web, which is that anyone should be able say anything they want
> >>>about existing resources [BERNERS-LEE98]. 
> >>>
> >>[BTW:  If nothing else, you've enabled me to spot a missing "to" in the
> >>sentence.  Thanks!]
> >>
> >>Specifically, the examples illustrate an independent party commenting on
> >>a description produced by someone else.  That (roughly) "anyone should
> >>be able to say anything about anything" is being cited as an
> >>architectural principle of the Web illustrated by the examples, not as a
> >>capability of RDF (and certainly not without exceptions).  I frankly
> >>find it hard to see anyone reading this, and being led to believe that
> >>they can then make RDF statements that violate RDF/XML (e.g., using
> >>rdf:ID as a predicate was one of the examples that came up in your
> >>interaction with Brian).  Of course, being able to "say anything about
> >>anything" provides the ability to state lies and nonsense (e.g., you
> >>could say rdf:ID rdf:type ex:MooCow), but we can't really prevent that. 
> >>Do you see a particular clarification that would help, such as a caveat
> >>here that there are technical limits to RDF's ability to express things?
> >>
> >>--Frank
> >>
> > 
> > Either RDF is able to ``say anything [...] about existing
> > resources'', in which case the sentence is germane, or it isn't, in which
> > case the sentence is irrelevant or misleading.  If RDF doesn't satisfy the
> > principle then there is no reason to keep the allusion.
> > 
> > I find this a general problem with the RDF documents.  A lofty principle is
> > stated, such as ``say anything ...'' or ``expressing information
> > ... without loss of meaning'', but RDF doesn't even come close to the
> > principle.
> > 
> 
> 
> Peter--
> 
> I understand your concern, but at the same time what you suggest ("If 
> RDF doesn't satisfy the principle then there is no reason to keep the 
> allusion") seems awfully black and white.  

I had thought that specifications, and the Primer is part of the RDF
specification, if only an informative one, were about ``black and white''.
If someone wants to write a paper on RDF, then it might be reasonable to
make these sorts of allusions, but I just don't see their place in an
official document on RDF.

> It seems to me reasonable to 
> be able to state the principle, even when we can't totally achieve it 
> (and of course, RDF is not alone in not coming close to this principle 
> in reality).   I agree, though, that an *uncaveated* statement of this 
> principle could be misleading.  What about adding the caveat that I 
> mentioned, that there that there are technical limits to RDF's ability 
> to express things?
> 
> --Frank

If the RDF documents were not liberally sprinkled with this sort of
overreaching allusions, then I might be satisfied with this approach.
However, fixing one instance of this general problem is not going to
eliminate the impression that RDF is much more than it actually is.

peter

Received on Monday, 17 February 2003 10:18:05 UTC