Re: SUO: RE: article on the pitfalls of metadata
On Mon, 25 Aug 2003, Richard Cooper wrote:
> > If you read Peirce's early work (1865-1870), you can see that
> > he began by
> > trying to figure out how science works, more broadly, to
> > understand the
> > logic of any method that deserves to be called scientific, in whatever
> > field that it might be applied. It was in order to do this that he
> > was forced to develop the theory of signs, or "semiotics", and the
> > logic of relative terms, a fragment of which was a major source
> > of predicate calculus and founded relational database theory.
>
> I see. I've been hearing about Peirce on this list for quite
> some time, but he always sounded more like a philosopher than
> a student of discovery processes.
He became the former due to his profound drive to study the latter.
> > In critiquing the Cartesian and Kantian models of science,
> > Peirce revived some old ideas of Aristotle that identified
> > three basic types of reasoning, that have come to be called
> > abductive, deductive, and inductive.
>
> >Abductive reasoning is
> > the form of reasoning that generates hypotheses, and thus it
> > is critical to the whole process of discovery and invention
> > in science. The nature of hypothesis formation, which is
> > also akin to diagnostic reasoning, is a thing that needs
> > to be better understood, but most people eventually come
> > to the conclusion that it is the most difficult kind of
> > reasoning to formalize, much less automate. About the
> > best you can do is clarify the conditions under which
> > it is apt to succeed, like restricting hypotheses to
> > consequential and falsifiable statements, that are
> > subject to deductive follow-up and inductive test.
>
> This sort of hypothesis formation is what many current
> discovery systems are using.
We're currently working on implementing it in Cyc through a "Scenario
Generation" project, which I'm part of.
And, to John, this is providing quite a testing ground for the upper and
middle ontology, exposing faulty / missing ontological work in the past.
(Why have we generated the hypothesis that event X happened in a
rest-room?? - this is a real case) and has resulted in a number of fixes /
improvements. So your grand claim about our developing our ontology in
complete isolation from unanticipated testing is just not true.
> It would be nice to read
> a formalization of it.
I'm not sure that a (complete) formalization is possible. Here is a
relevant quote, taken from an online dictionary of Peirce's terms at:
http://www.helsinki.fi/science/commens/dictionary.html
"Abduction is the process of forming an explanatory hypothesis. It is
the only logical operation which introduces any new idea; for induction
does nothing but determine a value, and deduction merely evolves the
necessary consequences of a pure hypothesis.
Deduction proves that something must be; Induction shows that
something actually is operative; Abduction merely suggests that something
may be.
Its only justification is that from its suggestion deduction can
draw a prediction which can be tested by induction, and that, if we are
ever to learn anything or to understand phenomena at all, it must be by
abduction that this is to be brought about.
No reason whatsoever can be given for it, as far as I can discover;
and it needs no reason, since it merely offers suggestions." (Harvard
Lectures on Pragmatism, CP 5.171-172, 1903)
For an extended account of how abduction, deduction and induction fit
together in a cycle to form the scentific method, see the (to me, fascinating)
paper, "On the Logic of Drawing History from Ancient Documents, Especially
from Testimonies", where Peirce has a crack at Hume's famous argument
against miracles. It can be found in _The Essential Peirce_ vol 2, as
well as other places.
Cathy.
--------------------------------------------------------------------------
Cathy Legg, Phd Cycorp, Inc.
Ontologist 3721 Executive Center Dr., ste 100
www.cyc.com Austin, TX 78731-1615
download OpenCyc at http://www.opencyc.org
--------------------------------------------------------------------------