W3C home > Mailing lists > Public > www-archive@w3.org > July 2001

[Fwd: context questions with public archive? [was:existential/conjunctive semantics (fwd)]]

From: Dan Connolly <connolly@w3.org>
Date: Tue, 31 Jul 2001 16:57:52 -0500
Message-ID: <3B6729E0.193668F0@w3.org>
To: www-archive@w3.org
forwarded with permission...

-- 
Dan Connolly, W3C http://www.w3.org/People/Connolly/

attached mail follows:


>pat hayes wrote:
>[...]
>Pat, I said a minute ago that I agree with all this stuff...
>that's mostly true, but actually, I don't understand
>some of it, but I'd like to... especially the
>bits about (1) quantifiers changing and (2) contexts
>and referential transparency.
>
>I'd like to ask you some technical questions, copying
>this material and my questions to www-archive@w3.org
>(which makes it available to the public),
>so that I can link to it from http://www.w3.org/DesignIssues/Logic
>and such.
>
>Is that ok?

Sure, though I will be off-mail until early next week.

>
> > >There's a Semantic-Web no-no for you: assuming
> > >too much about the universe of discourse.
> >
> > Its impossible NOT to assume a lot about the universe of discourse.
> > There isn't one all-purpose universe of discourse. If A writes
> > 'forall ?x....' and is thinking about everything they can possibly
> > imagine, B might still be thinking about something that isnt in A's
> > universe. For example (a real one) , A might be an endurantist who
> > thinks of physical things as retaining their identity through time,
> > whereas B might be using the EPISTLE model which describes everything
> > physical as 4-dimensional and talks about time-slices. You can
> > translate A's stuff into B's framework, but it takes some clever work
> > to do it; and when you do, the quantifiers change. If you just
> > conjoin A's axioms with B's and understand A's 'forall' in B's way,
> > you will quickly get inconsistency. If you give an informal English
> > gloss on what they are saying, they will seem to be saying the same
> > thing.
> >
> > > > What I don't understand how to think about is how this 
>relates to the Web
> > > > notion of a single space of identifiers (URIs) shared across contexts,
> > > > applications and ontologies. How much do we need to agree on across
> > > > all RDF/SW applications?
> > >
> > >We need to agree on a proof-checking algorithm,
> > >and we need a healthy marketplace of ontologies/axioms.
> >
> > We don't need to agree on a proof-checking algorithm. All we need to
> > agree on at the logical level is a proof theory or a model theory,
> > and then you can use one proof-checking algorithm and I can use a
> > different one.
> >
> > I agree about the market place. However, what is the kind of
> > discourse through which this market is conducted? (See my DAML
> > proposal for some thoughts on this, BTW.)
> >
> > >There are some interesting questions about co-reference
> > >or referential transparency or whatever
> > >in the proof-checking algorithm, meanwhile. e.g. CycL
> > >is not referentially transparent; a constant symbol (URI)
> > >can denote one thing in one context and something else
> > >in another.
> >
> > Yes, but since the contexts are explicit, the overall logic is
> > referentially transparent. Beware of that 'context' word, it means
> > different things to everyone. In Cycl, 'contexts' are Guha/McCarthy
> > contexts, ie first-class entities in the domain of quantification.
> > They are internal to an ontology, not external contexts of ontology
> > useage. The basic logic of this kind of context isnt much different
> > from good old traditional FOL; you could axiomatise it in KIF 3.
> > (BTW, both Guha and McCarthy make a basic mistake, IMO, by confusing
> > contexts of reference with contexts of inference. If you keep these
> > ideas separate, the logic actually gets easier. You have to stop
> > thinking of a context as something that the reasoner 'goes into' and
> > 'comes out of' , and instead just say that contexts are parameters
> > that allow a single name to have a variety of related, but distinct,
> > meanings. )
> >
> > ......
> >
> > > > Should the "Semantic Web" have a single answer to questions about
> > > > existence and time such as: can an object exist for a while then stop
> > > > existing then exist some more?
> > >
> > >Absolutely not; lots of answers are sensible.
> > >
> > > > I doubt we'd find consensus achievable, and
> > > > I worry (OK, I'm a worrier...) about the prospects for 
>interop amonst Web
> > > > apps that take differing views about such fundamentals.
> > >
> > >Yes, you worry too much. ;-)
> >
> > I think he is right to worry about this kind of issue.  Not that is
> > any way of getting it 'right', but mismatches at this level do create
> > huge interoperability problems in practice.
> >
> > >The way to get the Semantic Web to work is *not* to
> > >constrain folks to agree about real-world fundamentals, but
> > >to offer some valuable tools and ontologies and
> > >get the marketplace going.
> >
> > I agree. But the market is going to need something more than just a
> > basic logic.
> >
> > Pat
> >
> > ---------------------------------------------------------------------
> > IHMC                                    (850)434 8903   home
> > 40 South Alcaniz St.                    (850)202 4416   office
> > Pensacola,  FL 32501                    (850)202 4440   fax
> > phayes@ai.uwf.edu
> > http://www.coginst.uwf.edu/~phayes
>
>--
>Dan Connolly, W3C http://www.w3.org/People/Connolly/
>office: tel:+1-913-491-0501
>pager: mailto:connolly.pager@w3.org
>  (put return phone number in from/subject)

---------------------------------------------------------------------
IHMC					(850)434 8903   home
40 South Alcaniz St.			(850)202 4416   office
Pensacola,  FL 32501			(850)202 4440   fax
phayes@ai.uwf.edu 
http://www.coginst.uwf.edu/~phayes
Received on Tuesday, 31 July 2001 17:57:50 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 7 January 2015 14:42:01 UTC