'weasel wording' was Re: Difference between syntactic building blocks and formal languages ...

Peter F. Patel-Schneider wrote:
>
> > Re:  http://lists.w3.org/Archives/Public/www-webont-wg/2001Dec/0173.html
> >
> > Quoting from: Ian Horrocks (horrocks@cs.man.ac.uk)
> >
> > [[
> > These sorts of problem illustrates just why we need a precisely
> > defined semantics for our languages. If we allow for two possible
> > interpretations we may get into all sorts of difficulties:
> >
> > - it may be impossible to say something in DAML+OIL without stating
> >   something unintended in RDF (and vice versa)
> >
> > - it may be impossible to know which of two (possibly conflicting)
> >   meanings is the intended one
> > ]]
> >
> > But if we consider RDF\triples to be just syntactic building blocks with
no
> > formal semantics whatsoever, then would we still have this problem?
This
> > view would mean that the entailments of any statement are only the
> > entailments that can be infered by the axioms related to the arc label.
In
> > other words property arcs have semantics and entailments, but languages
like
> > DMLS, RDFS don't.  This has the advantage of allowing us to mix and
match
> > all the available properties of all the schema written in or
translatable
> > into NTriples.
>
> RDF does have at least an intended meaning, and is probably getting a
> formal semantics.  In light of this, I feel that any use of RDF triples
has
> to conform with the intended meaning.
>
> > Would that work?  If not, why not?
>
> I think that the problem is what a web ontology language syntax would want
> to use the parts of RDF that have a non-trivial intended meaning which
> would get in the way.
>

Pat Hayes' formal semantics for RDF describes as "weasel worded" RDF triples
as "asserted" under common conditions. The weasel wording leaving the
specific possibility that collections of RDF triples (aka a 'graph') may
_not_ be asserted, rather useful in formulae. I think this possibility is
the single most important part of the RDF meaning, leaving open the real
possibility that a language (WOL) can itself define what collections of
triples are intended to be asserted and which are intended to be used to
construct a formula (i.e. the formula is itself asserted without asserting
each triple that it contains).

Thus, WOL can probably use/be compatible with RDF-MT in some fashion, though
it may not choose to employ the RDF syntax (which does not currently have a
representation for triples that are not members of the set of asserted
triples).

Perhaps it should be stated more explicitly in RDF-MT, get rid of the
"weasel wording" and call a statement a statement.


Jonathan

Received on Wednesday, 2 January 2002 12:41:57 UTC