Dear Chris' M and W... (01)
Now look what you've done! You've drawn me into this! I'll get you
both for this! (02)
On Jan 27, 2007, at 22:10 , Christopher Menzel wrote: (03)
>> Accepting ontology == logical theory trivializes ontology, IMHO.
>
> Yes, it trivializes the *word* in order to get folks to focus on
> *quality* -- it's so much more productive to focus on what's good/bad/
> right/wrong about an ontology than to try to figure out some way of
> throwing a semantic fence around JUST the right logical theories that
> deserve the honorific title. Dude, it's a waste of time! (04)
Saying that it trivializes the *word* isn't really accurate since
what Chris W intends when he uses the word is a property, had by some
logical theories and not had by others. I agree that it is
*productive* to focus discussion on the nature of logical theories
used for the purposes people usually have in mind when they say the
word "ontology", but doing that doesn't replace a discussion of what
about those theories makes them useful for those purposes. (05)
That's why I sent my mail of a couple days back in response to Steve
Ray's comment about the focus of this discussion. IMHO, and in that
note, nobody gets to come to the table unless they acknowledge that
ontologies are logical theories. This seemingly trivial requirement
is one I've heard disputed in the past by, e.g., adherents of topic
maps. The point was simple -- if you want to say that your
"ontology" O has some useful role in guiding the action of software
to carry out some kind of function, then that role can be
characterized as a logical theory. Again, a seemingly trivial point,
but I'm trying to either silence those who think logic doesn't matter
or get them to admit that it does. Ok, so we're talking about logic. (06)
Now back to the purposes / functions I mentioned above. Just what
are those? What makes some logical theories "good" for the purposes
to which we'd like ontologies to serve and some bad. We heard a few
days back a report from one user of Dublin Core that it was "bad" (my
word). Why is it bad? I'm not sure I have a definitive answer, but
perhaps one I could offer is that it describes a fictional world --
one that serves some limited application purpose, but no more. That
is, it breaks when confronted with cases that many would consider
reasonable. (07)
>
>> It is exactly the
>> kind of thing I find myself fighting against time and time again.
>> It is the reason
>> the FOIS conference exists. I think its worth it to try and come
>> up with something
>> better.
>
> Really, Chris, is THAT what you're fighting? Aren't you fighting
> *bad* ontologies and *bad* ontological engineering? Give folks the
> word and teach them how to create *good* ontologies -- by example, by
> isolating general, qualitative principles of good practice, etc. (08)
But what could those be, Chris, if not *extralogical* principles? If
we wish to exclude any such principles, then all we're left with is
the rather cold machinery of logic. We understand, for a broad range
of logical systems under consideration for ontology engineering, what
their meta-theoretic properties are - we know what's definable &
what's not and lots of other things. We know what a model is and
what's not. So, what part of that can we modulate by isolating
general, qualitative principles of good practice? None that I can
see. One could attempt to borrow some principles from, say, software
engineering -- we can place requirements on documentation
completeness and the process used to arrive at this or that logical
theory. But those are hardly logical also. (09)
>> When writing standards, you can make use of words like "must"
>> "should" and "may" to
>> convey less-than-perfect definition.
>
> But the problem, by my admittedly dim lights, is that you won't get
> anything like definition at all. The idea of writing a standard for
> a notion of "ontology" that builds in vaguely specified qualitative
> features is like the idea of writing a standard for composing good
> music. It's a category mistake; we're not talking about a definable,
> standardizable thing here (at least, not in any useful sense).
> Standards can and should be written for things like logical languages
> and musical notation. Beyond that it's a matter of teaching/learning
> good practice. People start out writing bad music, but it's music
> all the same. (010)
I don't think this analogy hits the mark. Here your likening
ontology to music, or by your standard, likening logic to music. But
I think that misses precisely because the analogy should be between
logic and sound. Any sound (say one that humans can hear) is a
"good" sound just as any consistent logical theory is a "good"
theory. Now, one can argue about whether music is *just* sound but I
don't think anyone who's heard a crying baby or an explosion would
call that music. Music, supposedly causes the right things to happen
in the ear of the listener. It's sound that does something - not
just sound. (011)
So it is with ontology. An ontology is a logical theory that has the
"right" property/ies. What those are I'm not sure (I have some
ideas) but I'm pretty sure they'd exclude most logical theories just
as they would reasonably exclude Dublin Core. (012)
.bill (013)
_________________________________________________________________
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/OntologySummit2007/
Community Wiki: http://ontolog.cim3.net/cgi-bin/wiki.pl?OntologySummit2007
Community Portal: http://ontolog.cim3.net/ (014)
|