SUO: *Date 02 Apr 2002 -- Survey
¤~~~~~~~~~¤~~~~~~~~~¤~~~~~~~~~¤~~~~~~~~~¤~~~~~~~~~¤
Survey of Available Ontologies, Taxonomies, Thesauri, Lexicons, Logics
BA = Bill Andersen
JA = Jon Awbrey
JA: You have just summarized, more succinctly than I myself could ever hope
to match, quite obviously, all of the problems that I see in the way that
FOL and KIF are currently used -- that they are in fact deployed more like
implicit ontologies and folk weltanschauungen than genuine logics. The only
difference, I guess, is this -- that the users of FOL and KIF do not have
the same mass of specs before their eyes to blame for their confusion.
Now I know that many will still find this incomprehensible, but at
least now it's been clearly said. You are not just ipsing dixi.
BA: I think you're being a little hard on relatively benign creatures like
FOL and KIF, which is simply an alternate syntax for FOL. Now, you can
argue with the set-theoretic basis of their model theories if you like.
In fact, Barry Smith has done this on occasion. But if you don't take
the model theories *seriously*, but only as a modeling device (no pun
intended) you can say pretty much anything you want in them.
Yes, I have seen that bumper sticker:
"Syntaxes don't kill people. Taxes kill people."
So I would not want to be pulled into the fetishism of blaming dead syntaxes
for the poor workmanship of this or that concept-and-symbol-tool-user. Still,
what is "formalism engineering" if not the critique of existing formalisms and
the art of designing them more to the point of their transcendental purposes?
All of which is why I am trying to be careful, and to
address "the way that FOL and KIF are currently used",
rather than trying to guess at their possibly untapped
potential, especially if opened up to the environment
and allowed to grow, develop, and evolve. But I do
not see that happening yet, not around here, anyway.
So for now, it's back to the sorry state of things entire.
And here I suspect that the truth is something like the way
that it is with other tools, that different designs can make
different tasks differentially easier or tougher, and to the
point where some things that we would like to achieve become
darn near impossible, for all practical purposes, if tried in
the wrong sort of media.
Let me try to stick with this amazing but concrete observation
that just occurred: that you were able to see the mote in the
eyes of another linguistic community, to recognize that their
usage of a certain formalism embodies a host of ontological
assumptions, writs that you were able, from your outsider's
perspective, to view with no small measure of suspicion,
and would scarcely countenance taking for granted, and
even if, as you suspected out loud, they themselves
would probably fail to suspect the degree of their
loom's warp out of ontologically beige terrytory.
BA: Your recent posts have been filled with all kinds of mathematics,
all of which is based in set theory. Your earlier posts contain
all kinds of mathematical bits, all of which could be expressed
in first-order languages.
And all of which could be expressed in roman numeral gödel codes,
if you like that sort of talk. But you will notice that my chosen
authors chose to express themselves rather differently. You will
notice also that their theorem-to-axiom ratio is somewhat higher
than that which seems to be local batting average.
BA: Bottom line is that FOL variants aren't the problem.
I knew _|_ ...
_|_ was a Friend of Mind ...
And _|_ the Weaver says Other Wise ...
BA: The problem is with languages like OIL which are touted (even
more loudly by those who don't design them) as silver bullets
for ontology. Having recently read some of the OIL stuff I
was shocked at how restrictive it is and how it forces bad
philosophical choices on you -- no matter what your
philosophical stand is.
And now you know how I feel about FOL and KIF, in their local avatars.
BA: With respect to FOL -- I think you're going to be hard pressed
to find some *computation* that you cannot express in FOL.
The same thing goes for doing calculus, after the fact, in roman numerals.
This is to confound the dynamics of discovery with the routines of report.
BA: After all, we're talking about implementing
ontologies to run on computers here.
Exactamundo!
There are ways of living with a formalism where one is barely aware
of it as a formalism, where one glosses over its signlike character,
much less stops to refract on the physical embodiments of its syntax.
I lived in so blithe an idyll -- how I wish that I could now go back!
But I remember the tumult that the concepts first of information and
then of computation brought into my life, where I had to become a bit
painfully self-conscious about the limitations of our human finitude
in the struggle upstream against the onrushing cascades of entropy.
In time I discovered that beginnings of this sea change in how we
weigh signs, in the light of their informational and computational
properties, is a fairly new invention. Now, it seems that ancient
writers were always somewhat more aware than we have been of late
that logic and signs were bound up with each other, though Locke
gets the credit for consummately minting the coin commemorative
of that fact, but it was not until the 19th Century that a truly
new idea comes into play, that signs have limits in what they can
signify, and that our ability to process them is likewise strained.
Peirce built these ideas into his first logic systems, but he was
the last logician to understand science and its modes of inquiry.
And the systems that we find most commonly used hereabouts today
go gadding about like aristocrats as if this revolution had never
happened -- can you imagine? -- as if they still had their heads!
Jon Awbrey
¤~~~~~~~~~¤~~~~~~~~~¤~~~~~~~~~¤~~~~~~~~~¤~~~~~~~~~¤