ONT Re: Abductive Knowledge Acquisition (AKA)
¤~~~~~~~~~¤~~~~~~~~~¤~~~~~~~~~¤~~~~~~~~~¤~~~~~~~~~¤
Jon Awbrey wrote:
>
> ¤~~~~~~~~~¤~~~~~~~~~¤~~~~~~~~~¤~~~~~~~~~¤~~~~~~~~~¤
>
> Randall, Doug, & Fellow Abductees,
>
> I have changed the title to protect our in-a-sense under this alias.
>
> I found a half a reply to each of your notes in my draft folder,
> from which evidence I abduce my intent to say something more on
> the topic -- give me a moment while I do a bit of revisioning
> of my personal intellectual history -- way last week! -- and
> one that I already barely can recall.
>
> By way of stalling, while my ITM (intermediate term memory) kicks in,
> let me give you a bit of my personal history with this subject area,
> as I see it, the place of abductive reasoning within inquiry, whether
> it be ontological, phylologicial, or any other community or direction
> of inquiry that is at stake at the moment.
>
> I more or less kept up with the literatures on abductive reasoning
> and inquiry in general in phylosophy (it recapitulates ontosophy)
> and in both CS's, cognitive science and computer science, up until
> about 1994, when I sequestered myself to begin dedicated work on
> my dissertation. This is relative to a starting point in 1969,
> when I began my simple desultory poking around in the n'hood,
> according to the fashion of one's undergraduate Wanderjahren.
>
> Except for the occasional pieces of exceptional work that
> I mention later, the typical paper in this area, at best,
> begins with a token nod in the direction of Aristotle's
> having given the subject a name, namely, "apagoge", and
> to Peirce's having given it a local habitation within
> his "theory of inquiry" (TOI), at which point it will
> cite one of Peirce's most partial explanations and/or
> popular expositions of what abductive inference is or
> ought to be, and, with this grudging devoir to the dim
> ancestry and to the even dimmer ancestors of the subject,
> it will promptly launch into one of those "whistling in the
> dark to stave of the 'horror of history' (HOH) pot-boilers",
> so popular among the genres of treatises that take up the
> usual suspect stance and make up the staple mutilated
> diet of our late grate modern "scientific" attitude:
>
> | But, of course, from our presently evolved state of knowledge
> | we all know that all this stuff, so dimly presaged by even the
> | greatest minds of those primitive times and rudimentary mores,
> | comes down to this minor aspect of our Old Familiar Friend,
> | <elect one> Deductive Mill-Grinding, or Bayesian Inference.
> | The End.
>
> Well now, I obviously need that second cup of coffee,
> as I am down to the all too bitter dregs of this one.
>
> More, Later,
>
> Jon Awbrey
>
> ¤~~~~~~~~~¤~~~~~~~~~¤~~~~~~~~~¤~~~~~~~~~¤~~~~~~~~~¤
To continue, besides all of my grousing and other displays
of ruffled feathers there is actually a substantive point,
which I will ever endeavor to render more clear. This is
more important than the acclaim or the reclaim of a given
gifted thinker, as I believe that Peirce was -- it has to
do with whether, in word or in deed, we are making use of
the best available ideas and information with respect to
any intellectual or practical effort that might be named,
and where are the mortemain blocks in the way of inquiry,
the obstacles and the resistances to our doing just this.
Peirce is often recognized as "that guy who discovered all kinds of
stuff 30 or 50 years ahead of anybody else, but who did not publish
very much of it, and so a bevy of other folks had to come along and
re-discover it and render it into public knowledge for our benefit".
This is baloney. Much of what he discovered about logic and about
what he called the "theory of information" was presented in public
lectures at the "epicenters" of research in those days -- and I use
this earthshaking word "epicenter" in a precise, technical, if still
metaphorical sense -- at least as far as the frontier science on the
colonial scene was concerned, or else published in the A-List Journals
of the Day.
For the extent to which it was actually Peirce who poured the
concrete for the foundations of our present logical residence,
syntactically speaking, at least, you may read Hilary Putnam's
essay on the subject -- sorry, I will have to dig up that cite
a little bit later.
But wait, it gets worse. It is closer to the truth to say that
many of the most important insights that Peirce had and uttered,
even those that were solidly embodied or starkly implicit in his
earliest lectures and publications, have yet to be re-discovered,
or else were actually dropped, forgotten, or ignored by the later
adopters of his precocious brain-bairns.
And again, the important thing, from the standpoint of our
not yet enlightened self-interest, is that not a few really
effective and powerful ideas have yet to see a general notice
that their true time has come. The time has long past arrived,
but we have yet to catch up to it.
Well, this sort of diatribe makes me as weary as it no doubt does you,
but then, I am not really intuit for my own particular private health.
Later, & Anon,
Jon Awbrey
¤~~~~~~~~~¤~~~~~~~~~¤~~~~~~~~~¤~~~~~~~~~¤~~~~~~~~~¤
> Remainder of initial discussion follows:
>
> Randall R Schulz wrote:
> >
> > Doug, Jon & Others WIMC,
> >
> > It sounds like there's some conceptual overlap between
> > your methodology and the kind of automated support for
> > knowledge acquisition whose creation is the object of
> > the current DARPA RKF Project.
> >
> > Perhaps perusal of these resources will prove interesting:
> >
> > - DARPA Official RKF Page:
> > http://dtsn.darpa.mil/iso/index2.asp?mode=9
> >
> > - RKF Contractor Home Page:
> > http://reliant.teknowledge.com/RKF/
> >
> > - RKF KRAKEN
> > (Knowledge-Rich Acquisition of Knowledge from Experts who are Non-logicians)
> > Project Home Page:
> >
> > http://reliant.teknowledge.com/RKF/projects/KRAKEN.htm
> >
> > - KRAKEN Project Proposal:
> > http://reliant.teknowledge.com/RKF/proposals/Cycorp.html
> >
> > - GMU RKF Home Page:
> > http://lalab.gmu.edu/RKF/default.htm
> >
> > - GMU RKF Proposal:
> > http://lalab.gmu.edu/RKF/proposal/default.htm
> >
> > - SRI's RKF Home Page:
> > http://www.ai.sri.com/~rkf/
> >
> > - Other SRI RKF Links:
> > http://www.ai.sri.com/~rkf/links.html
> >
> > - ISI RKF EXPECT Page:
> > http://www.isi.edu/expect/rkf/
> >
> > - UWF (Univ. of West Florida) Institute for Human & Machine Cognition RKF Page:
> > http://www.coginst.uwf.edu/projects/rkf/
> >
> > - Stanford RKF Page:
> > http://www.ksl.stanford.edu/projects/RKF/
> >
> > By the way, although as you can see Teknowledge
> > is participating in this project, I found these
> > resources via a Google search for "RKF" and for
> > "RKF and KRAKEN". (I omitted the page on the
> > Kennedy assassination that came up because
> > someone mistyped RFK ...)
> >
> > Randall Schulz
> > Teknowledge Corp.
> > Palo Alto, CA USA
> >
> > At 07:23 12/3/2000, Douglas McDavid wrote:
> > >
> > > Jon --
> > >
> > > In your reply to Jim you write:
> > >
> > > > It would take a whole lot of work to transform the current version of KIF
> > > > into the sort of representation that would be useful as a discovery tool.
> > > > And I have not noticed anybody on that side of it being even the least
> > > > bit open to suggestions in that direction.
> > >
> > > When I think of a discovery tool I actually think of a discovery process,
> > > supported by tool(s). This process goes something like this:
> > >
> > > 1. Identify a domain of interest.
> > > 2. Assemble a corpus of representative material from that domain.
> > > 3. Parse that material to extract terms.
> > > a. Remove "stop words".
> > > b. Extract lexical items (single words and
> > > relatively indivisible word groupings).
> > > 4. Determine which of these lexical items matches a known meaning.
> > > a. Check existing ontolog(ies) for existence.
> > > b. Make a judgment call with respect to sameness of meaning.
> > > c. Record usage of this common term for the new
> > > domain of interest ontology (DOIO).
> > > 5. Create ontological structure for new terms.
> > > a. Determine (define) precise meaning of new term
> > > in the context of usage.
> > > b. Isolate meaning components.
> > > c. Determine correspondence between meaning components
> > > and existing ontological elements.
> > > d. Record componential correspondence.
> > > e. Create any new ontological elements required by
> > > this domain corpus.
> > > f. Record new elements, in relation to this DOIO.
> > >
> > > etc.
> > >
> > > I tried to keep that process description generic.
> > > I envision several software components that can benefit the work,
> > > but I can also envision it completely as a manual process using
> > > paper documents and 3x5 cards. I would prefer to have a good
> > > parsing engine, a word processor, and a very flexible,
> > > multidimensional sort of database. I'm coming to
> > > appreciate the place of logical axiomitization
> > > in such an architecture.
> > >
> > > I'm trying to get to the "rudiments" (as in rimshot and paradiddle)
> > > of this kind of work. Is this at all what you have in mind when you
> > > evaluate representations and tools from a discovery perspective?
> > >
> > > Doug McDavid
> > > Certified Executive Consultant
> > > Business Innovation Services - IBM, US
> > > Member of IBM Academy of Technology
> > > mcdavid@us.ibm.com -- 916-549-4600
>
> ¤~~~~~~~~~¤~~~~~~~~~¤~~~~~~~~~¤~~~~~~~~~¤~~~~~~~~~¤
¤~~~~~~~~~¤~~~~~~~~~¤~~~~~~~~~¤~~~~~~~~~¤~~~~~~~~~¤