Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

Re: SUO: Vote 2001-02: IFF Foundation Ontology




Robert,
   Thanks for your replies.  My further comments and questions below:

At 04:53 PM 8/20/2001 -0700, Robert E. Kent wrote:

>Adam,
>
>(see the comments below)
>
>----- Original Message -----
>From: "Adam Pease" <apease@ks.teknowledge.com>
>To: "Robert E. Kent" <rekent@ontologos.org>
>Cc: "SUO" <standard-upper-ontology@ieee.org>
>Sent: Monday, August 20, 2001 3:03 PM
>Subject: Re: SUO: Vote 2001-02: IFF Foundation Ontology
>
>
> > Robert,
> >    I appreciate the very concrete message.  This is exactly the sort of
> > discussion needed if we're to create an actual standard.
> >
> > 1.  I notice that you're using sorted quantifiers.  That isn't in SUO-KIF
> > nor the latest KIF group draft that Michael Gruninger has prepared,
> > although they are in the version at
> > <http://logic.stanford.edu/kif/dpans.html#6.3>.   You don't appear to cite
> > which version of KIF you're using in your IJCAI paper or proposal for
> > SUO.  Which version do you use?
>
>Comment #1:
>I am anticipating what will be in the new KIF. See Chris Menzel's comment
>
> > Looks good to me.  I like your restricted quantifiers (and descriptions)
> > but the syntax for them is not settled.  Most likely it will be
> > something like (<quant> (<var>+ <sentence>*) <sentence>).
>
>in a personal message to me. These restricted quantifiers can be taken as
>simple abbreviations with the obvious expansions: conjoining for the
>existential quantifiers and anteceding in an implication for the universal
>quantifier.

thanks for the clarification

> > 2.  You've shown below how a natural language sentence could be expressed
> > in a way consistent with a version of KIF and the IFF.  However, it's not
> > clear (to me at least) what value is being added by IFF itself.
>
>Comment #2:
>The IFF Foundation Ontology defines a rather elaborate semantic mechanism.
>This example gives an indication of how to translate an ontology expressed
>in the new KIF to the very precise terminology of the IFF Model Theory
>(sub)Ontology (to be released in version 3.0 of the IFF Foundation
>Ontology). Although there are many IFF statements needed to represent the
>KIF statement, these IFF statements are *extremely simple* (there are
>neither quantifiers nor logical connectives, just membership assertion and
>function application). As a result, the IFF representation can easily be
>represented in standard programming languages.

Could you show an example representation in KIF and then in Java or C or 
some other common language?  This seems to me a very strong and non-obvious 
claim.  Expressing the semantics of any logical language in a procedural 
programming language seems rather difficult to me.


>Once the KIF ontology is
>internalized in IFF, it then corresponds to one particular formal concept in
>the truth concept lattice indexed by that ontology's 1st-order language.
>Meets and joins in this lattice allow us to automatically combine this
>ontology with any collection of ontologies. For example, if the SUMO
>were resolved into several meaningful component ontologies, and these were
>individually translated to the internal IFF representation, then the meet in
>the truth concept lattice would correspond to the IFF representation of the
>whole SUMO. Note, a collection of ontologies are inconsistent when their
>meet is the bottom formal concept.

How would this different from simply concatenating two KIF files and 
checking for a contradiction?  How does IFF itself support a merge?  As a 
concrete example, let's say we have

------
file #1
   (instance Adam Person)
------
file #2
   (instance Bill Human)

In this trivial case one concatenates the two ontologies and it's a merge 
but a poor one.  There's no way to know that Human and Person are 
equivalent unless a human being looks at the two files and makes an 
educated guess.  One can imagine more complex cases but the issue for a 
merge is just this ultimately.

>  As a second benefit, once an ontology is
>internalized in IFF, 1st-order interpretations allow us to translate between
>various truth concept lattices.

Could you explain this in more detail, maybe with an example?  I expect 
many folks don't know what a "truth concept lattice" is.

>As a third benefit, we have the opportunity
>to combine ontologies using colimits. This allows the summation and fusion
>(quotienting) of ontologies.

Could you provide explanation and an example for this as well.

>  As a fourth, more conjectural benefit, Murray
>Bent (Yang Yun) is interested in defining a lower metalevel module for game
>semantics. This could possibly correspond to the game-theoretical semantics
>discussed by John Sowa
>[http://users.bestweb.net/~sowa/misc/mathw.htm#Model].

That sounds like a useful effort but how is it an example of the utility of 
IFF?

> > You use
> > the term 'event' in the formalization below but no such term appears in
> > IFF.  In contrast, SUMO does have notions of events (and times, roles of
> > actors etc) which provide a definition for that term.
>
>Comment #3:
>I am assuming that this example is just a fragment within an ambient
>ontology,
>such as the SUMO, that has an "event' term.

Then I would take this either as simply a task that remains for someone to 
merge the SUMO content under the IFF structure, in order to provide the 
ontological content that is relevant for real-world application modeling, 
or as a disadvantage of IFF since it lacks this necessary content.

> > The problem (I would say) with this formulation as well as what you
>provide
> > below is that there is no upper ontology being used.  The term 'event' is
> > meaningless because it is just a token with no formal definition.  In
> > contrast, the SUMO formalization would be
>
>Same comment #3.
>
> > SUMO already has definitions for concepts like Eating, the stuff an object
> > is made of ('material') etc.  Instead of being just tokens, the concepts
> > have some meaning by virtue of their associated axioms.  That's why the
> > list of statements prior to the formalization of the English example
> > sentence is so small - all those concepts already exist in SUMO.
>
>Same comment #3.
>
> > So, in summary, what does the IFF version of the example provide us that
> > the SUO-KIF-only version does not?
>
>Same comment #2.:

Thanks again for the replies.  I still haven't understood the advantages of 
IFF but hopefully another round of explanation will get us closer.

ADam


>Robert E. Kent
>rekent@ontologos.org

Adam Pease
Teknowledge
(650) 424-0500 x571