[Top] [All Lists]

Re: [ontology-summit] First Model Bench Challenge

To: ontology-summit@xxxxxxxxxxxxxxxx
From: John F Sowa <sowa@xxxxxxxxxxx>
Date: Sun, 06 May 2012 10:37:56 -0400
Message-id: <4FA68CC4.5080102@xxxxxxxxxxx>
Doug, Adrian, Leo, and Jack,    (01)

> I thought that we were dealing with ontological issues, not NL issues.    (02)

> I'm not sure how one can legitimately claim to be studying Semantics
> without at least touching on NL.
> NL is difficult, it's complicated, but it's essential that the intended
> real world meaning behind p123(x,y,z) be captured computationally.    (03)

I recognize what Doug is trying to say, but I agree more with Adrian.
In fact, I would make even stronger claims:    (04)

  1. Natural languages are the ultimate knowledge representation
     languages -- in generality, flexibility, *and* precision.    (05)

  2. Every version of logic -- in fact every artificial notation
     of any kind -- is an abstraction from some aspect of some NL
     for some special purpose.    (06)

  3. For many purposes, the artificial languages can be more
     precise in what they say -- because they are more limited.    (07)

  4. That gain in precision can be of immense value for many
     purposes, as we see with computer programming languages.    (08)

  5. But that gain in precision is often illusory because it
     strips away the context and nuances of the original NL.    (09)

  6. Re formal logic:  What it says so precisely may be totally
     different from what the author had intended.    (010)

See http://www.jfsowa.com/pubs/rolelog.pdf    (011)

> However, semantics may also obscure the ontological analysis
> because we do catch on the ways to refer and not necessarily
> the referents.    (012)

I think you meant "semantic notations" rather than semantics.
In any case, I agree that many people confuse semantics with some
particular notation.  I also agree that *every notation*, including
NLs can cause that kind of confusion.  That's why it's important
to learn more than one NL.    (013)

My recommendation is to learn *multiple* notations and practice
translating from one to the other.  The patterns that remain
invariant through all the translations are far more fundamental
than the quirks of any particular version.    (014)

> We may have quite different labels for the same concepts
> so let me try some others...    (015)

Yes.  Metalanguage is necessary for any kind of discussions about
knowledge representation.  But it is essential to learn *multiple*
metalanguages in order to avoid getting hung up on quirky notations.    (016)

> Any suggestions about how to clean up my act?    (017)

In my Knowledge Representation book, I used two very different
metalanguages:  English and predicate calculus.  In that book,
I discussed many, many different notations, including conceptual
graphs.  But I always showed how to translate each notation to
and from English and predicate calculus.    (018)

Fundamental principle for teaching knowledge representation:
Always teach students three or more notations in addition
to their native language -- one of them should be predicate
calculus and one of them should be some kind of diagramming
notation, such as UML.    (019)

Make it very clear that the metalanguage for talking about
any notation is far less important than the patterns that
remain invariant across multiple notations.    (020)

John    (021)

Msg Archives: http://ontolog.cim3.net/forum/ontology-summit/   
Subscribe/Config: http://ontolog.cim3.net/mailman/listinfo/ontology-summit/  
Unsubscribe: mailto:ontology-summit-leave@xxxxxxxxxxxxxxxx
Community Files: http://ontolog.cim3.net/file/work/OntologySummit2012/
Community Wiki: http://ontolog.cim3.net/cgi-bin/wiki.pl?OntologySummit2012  
Community Portal: http://ontolog.cim3.net/wiki/     (022)
<Prev in Thread] Current Thread [Next in Thread>