SUO: RE: RE: Re: Architecture of an intelligent ontology development algorithm
-----Original Message-----
From: Richard Cooper [mailto:rich@valutech.com]
Sent: Wednesday, August 27, 2003 11:55 AM
To: Tom Johnston; Jon Awbrey; SUO
Subject: RE: RE: Re: Architecture of an intelligent ontology development
algorithm
Tom Johnston wrote:
>
> Or, perhaps even more generally, the "better question" is how
> to combine
> intentional, purposive design with natural selection
> operating on trial and
> error? I think R&D in pharmaceuticals is a good working
> example of that
> combination.
Agreed.
> As to an upper ontology, I do not think that waiting for
> natural selection,
> operating on a pool of candidates into which some process
> (analogous to
> genetic mutation) introduces novel candidates or
> candidate-components, is
> likely to succeed in any time frame meaningful to us. I think
> we need an
> intentional, purposive design, tested for adequacy at each
> stage of its
> evolution, against a wide range of low level ontologies taken
> from a diverse
> set of concerns represented by working databases in various areas of
> academic research and business function.
Most "genetic algorithm" implementations I've read of are
using combinations of lower level elements, which then become
newly available for use in yet more complex combinations.
So its not the terminals that are being combined directly, as
is supposed the case in natural selection.
But in natural selection, most mutations are thought to be the
result of cross-overs and copying errors rather than rare
changes of a nucleic acid. So natural selection builds on
top of existing successes also.
Studies of GA models shows that surprisingly few generations are
needed to make useful new results. So I disagree with the belief
that lots of time has to pass before meaningful results can be
obtained.
Abduction, as has been attributed to Peirce, is carried on
in a mechanical fashion by the cross-over and copying error
techniques in GA and genetic programming (GP) models, and
appears to work rather well there to focus the new terms in
a direction that is productive.
Just as Einstein had to build on top of Newton and Michaelson-Morley,
GA and GP functions build on their own intermediate successes.
Isn't this a form of abduction? New hypotheses are formed from
old ones and from recently successful ones ad infinitum.
TJ: Sounds like it to me. Comments like this are helping me get the hang of
"semiotic-talk". Thanks.
> Top-down (intentional, theoretical), constantly tested and refined by
> comparison with bottom-up (evolutionary, real world).
>
> The testing and refining never stops, of course. So in
> theory, it could lead
> to revisions in the highest levels of the ontology. Only
> through vacuousness
> could the highest levels of our ontologies become immune to
> revisionist
> pressures.
>
> This, of course, is pure Quinean holism. But although,
> according to Quine,
> even the laws of arithmetic are in principle subject to
> revision in the face
> of recalcitrant experience, we count on them as being pretty
> stable. And
> they have been. By the exact same token, I would expect a
> good upper level
> ontology, once proven stable against a couple of dozen large,
> robust and
> successful real world databases, to settle down into a stable state.
That I disagree with. Natural evolution hasn't stopped,
and we may now use intelligent selection methods to organize
new life forms for our own benefit, but I don't see a reason
to think it will ever become static, any more than our own
goals become static and stable.
TJ: I meant "relatively stable", of course, "stable" in the sense that Quine
thought arithmetic to be. That sense being that, if we were ever faced with
a serious empirical counter-example to the laws of arithmetic, (something in
the strange world of quantum physics, perhaps), we would find some other way
to accommodate that example than by revising the laws of arithmetic. Even
more likely, we would be unable to even describe the phenomenon in a way
that brings the laws of arithmetic into question. So I think an upper level
ontology is more like logic and arithmetic -- something near the core of
Quine's conceptual sphere -- than like "our own goals". Why do you think an
upper level ontology is as subjective, ideosyncratic and mutable as personal
goals are?
Here's why I think this particular issue is important. If an upper level
ontology can be or become as relatively stable as I have suggested, then it
can have legislative force, contra (apparently) to what John has suggested.
In other words, just as we do not think about revising arithmetic when we
encounter anomalies in counting-things kinds of activities, we can reach the
point where we do not think about revising an upper level ontology when we
encounter anomalies in classifying-things kinds of activities. Of course,
this isn't an absolete "not"; it's a Quinean "not".
To keep the upper level ontology stable, we will sometimes have to work
hard, fooling around with the middle level ontologies sometimes, and the
lower level ontologies often. In rejecting a lower level ontology that we
cannot fit into our upper levels, we are saying that we choose to describe
the empirical phenomena we are confronted with differently than it is
described by the suggestion we are rejecting. For example, talking about a
sick child, we may say to a witch doctor "You may choose to say that the
child is possessed by an evil spirit, but I say that what we have here is a
child infected by the xyz virus". The former way doesn't fit into our
ontology of medical conditions; the latter way does.
Now someone is likely to respond: "But the witch doctor can't cure the
child; modern medicine can". Well, sometimes the child gets better after the
witch doctor says his demon-defeating spell, and usually modern medicine can
do little better with a viral infection. Another example: the chemical
theory of phlogiston was never conclusively proven wrong. Scientists just
stopped trying to make it work. Another example: Kuhn made the same point
about the geocentric astronomical theory; with the use of such devices as
epicycles, it was able to "save the appearances", i.e. account for the
phenomena observed through that new scientific instrument, the telescope.
And who is to say that Chomsky succeeded because he discovered the "truth"
about language, and that consequently there are (ontologically speaking)
innate language acquisition structures in the human brain, there are
linguistic "deep structures" and so on? I think one reason for Chomsky's
success is that he provided a game that a lot of graduate students could
play, obtaining PhDs and earning a living in the process.
So I claim that an upper level ontology can be as highly stable as the laws
of arithmetic, but in the only way that such stability can ever be
achieved -- by us. We can MAKE it stable, by accommodating any purportedly
conflicting evidence by re-interpreting that evidence, i.e. by making
adjustments in our ontological schemes somewhere ELSE.
But if you are right that upper level ontologies are as ephemeral as
personal preferences, then our attempts to make a chosen one stable will
appear more and more artificial. Our upper level ontology will start to look
more and more like the views of the flat earth society. So let's try it and
see. I take it that that is what is happening with CYC, SUMO and Dolce. As a
Quinean, I do not look for a single acid test, conclusively falsifying each
of the bad ontologies. But I do look for these various candidates to
eventually spread out along a spectrum of conceptual comfort, one seeming
more natural, intuitive than another, another seeming more convoluted by
comparison. As this happens, researchers will find that one just "feels
right", a feeling which I interpret as a disposition to MAKE experience fit
into the chosen candidate.
Thus, I think, scientific theories (at least on the large scale) rise and
fall, as Kuhn claimed. Thus, I think, will progress in developing ontologies
be made, with an upper level ontology being an effort at categorization
certainly "on the large scale".
What do you think? First, do you think there is a real difference here, as I
think there is? Secondly, if you differ with my conclusions, as you appear
to do, where do you think I've gone wrong?
Tom
> Nor does the benefit flow in one direction only -- lower
> level ontologies
> helping with the development of higher level ones by being
> test cases for
> their applicability. Upper level ontologies can also help us
> develop better
> lower level ones, by revealing patterns in that lower level
> data that the
> originating "subject matter experts" had never seen. I
> provided a brief
> manufacturing example a week or two ago. Another set of
> examples come from
> generalizing from a set of relational tables (or OO classes)
> to a common
> supertype table (or class). Several vendor-provided "industry
> standard" data
> models, such as IBM's banking model, define an
> INTERESTED-PARTY relational
> table, subtypes of which include CUSTOMER, VENDOR, COMPETITOR,
> REGULATORY-AGENCY. (Of course, this doesn't amount to very much, since
> relational DBMSs support very little of the semantics of
> super/sub types. In
> fact, in relational databases, they come to nothing more than
> one-to-one
> relationships between the supertype and each subtype, optional for the
> supertype, required for the subtype. So although the data
> model diagram with
> its type hierarchy looks very sophisticated in a vendor's
> slide show, when
> it gets down to implementation in a working database, it's
> much ado about
> very little.)
Yes, the object structures in the design process get pressed
flat in the data model.
> Nonetheless, to summarize: it's top-down and bottom-up, design and
> trial-and-error. If I have any content to add to this truism,
> it's this: the
> process should be more top-down at the top, more bottom-up at
> the bottom,
> but always both, at all levels. The influence works both ways. We
> ontologists have something to add, something that reaches all
> the way down
> into insurance claims processing databases, transportation
> freight bill
> reconciliation databases, and grocery store shopping basket analysis
> databases.
Agreed.
-Rich
> In particular, as I argued in an earlier email about manufacturing
> databases, we should not think that the "real truth" is found in
> functioning, real world, bottom-level databases. The
> ontologies they embody
> are often confused, and the codebase and user knowledge of the system
> substantially devoted to compensating for the confused
> ontologies. The whole
> things are Rube Goldberg (Heath Robinson, for the Brits) contraptions,
> usually and for the most part.
>
> Tom
>
> -----Original Message-----
> From: owner-standard-upper-ontology@majordomo.ieee.org
> [mailto:owner-standard-upper-ontology@majordomo.ieee.org]On Behalf Of
> Jon Awbrey
> Sent: Monday, August 25, 2003 5:45 PM
> To: SUO
> Subject: SUO: Re: Architecture of an intelligent ontology development
> algorithm
>
>
>
> o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o
>
> [Reposting after 2 hours]
>
> Rich,
>
> I would have thought that a fairer summary of what's been said here
> all summer long on many different threads is that there is really no
> such thing as a hypothesis-free algorithm for discovery -- actually,
> that's more like a summary of what's been discovered about discovery
> over the last few thousand summers, but who's counting? So I think
> that a better question might be something along the following lines:
>
> How are concept-driven (analytic, axiomatic, rationalist,
> top-down) methods
> and data-driven (synthetic, contingent, empiricist,
> bottom-up) procedures
> best to be integrated in human inquiry, or in the
> reconstitutions thereof,
> given that the distinction between analytic and synthetic is more
> relational,
> interpretive, or "situated" than it is absolute, invariant,
> or "essential"?
>
> A start on answering that question might be to get a better
> analysis of the
> similarities among and the differences between the various types of
> reasoning
> that need to be integrated. On that score, my advice would
> be: Read more
> Peirce.
>
> Jon Awbrey
>
> o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o
>
> Richard Cooper wrote:
> >
> > Since many of us seem to agree that a bottom-up
> > algorithm could be used to produce the axiom
> > set of an ontology through situated experience
> > in the real world, I'm trying to draft some
> > requirements for this algorithm.
> >
> > There is a very suggestive paper at
> > http://jasss.soc.surrey.ac.uk/6/3/1.html
> > "Discrete Agent Simulations of the Effect
> > of Simple Social Structures on the Benefits
> > of Resource Sharing".
> >
> > The paper desribes a simulation of agents in an
> > environment somewhat like early human societies
> > are thought to have evolved in.
> >
> > A similar approach could be used to measure the success of
> > each strategy on the basis of how successful agents use that
> > strategy. In a simulated environment, instead of a situated
> > one, its easy to measure behaviors and organize them according
> > to what works well and what doesn't.
> >
> > So in a situated environment, perhaps the algorithm can guess at
> > axioms based on fragments of previous guesses that were successful.
> > The so-called evolutionary algorithms could suggest requirements for
> > monitoring the algorithm's behavior in the real world,
> measuring success
> > and failure, and buliding a database of experience for process
> improvement.
> >
> > So it seems to me that the process improvement concepts should be
> > a top level ontology in an algorithm that learns still higher level
> > axioms, while the WordNet concept set provides at least the
> words for
> > communicating with real world people.
> >
> > Any thoughts on this subject?
> >
> > Thanks,
> > Rich
>
> o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o
>
>