|
Gary,
What I'm looking for in all this is a way to
improve the level of automation
for natural language applications. So an
objective/subjective dichotomy
would be better if it could be measured, predicted,
calculated, and otherwise
algorithmically determined. It doesn't have
to be a validated measure, just
so long as it is useful.
Given such a measure, we could organize ambiguous
interpretations into
those that are more objective versus those that are
more subjective. That
ranking would be useful, presumably, in working on
utterances.
JMHO,
Rich
----- Original Message -----
Sent: Thursday, March 24, 2005 11:11
AM
Subject: RE: nature -> "human brain"
-> "language terms" ==>> knowledge ?
Rich,
I’m pressed to respond to your
point(s) to the degree it/they merits. But here are a few
thoughts.
RC >Nice description, but how
do you represent degrees of subjectivity? How would >you define a
gradient - number of hypotheses divided by (hypotheses plus "validated"
>facts)?
RC>One problem is that the
"validated" hypotheses of some people are what I would call subjective
errors. And they would think the same of my "validated"
hypotheses. So even the validation process is very subjective (or very
thirdish). So most people just treat it RC>like pornography - they
know it when they see it.
I can say that this aspect of the
thread started with Sowa arguing that we need to go beyond simple ideas of
Objective and Subjective and I agree with this, but can’t say that I have a
grand scheme to do this. The
example I offered attempted to explain a part of the differences as due to
mental models with varying degrees of empirical “validation”. It seems
plausible that there will many types and I don’t assume a simple
classification which may vary by type of problem. I wouldn’t look for a ratio
of Hyp. to Facts to do it and as you point out our assessment of these is
subjective itself. But I hope we
can do better than leaving it at “I know it when I see it.” We can agree on certain types of
standards that make one hypothesis better than another and elevate one to a
theory based on confirmations.
That’s the general path I’d take and it externalizes some of the
criteria by attaching them to the results of
observations.
Throughout this discussion I have
been speaking casually of hypotheses as cognitive states together with
processes as constituting organized information-bearing structures (aka
representations) of one kind or another.
However, this assumes that hypotheses as mental representation may be posited
as a mental objects with certain semantic properties (content, reference,
truth-conditions,
truth-value, etc.) that allow easy computation. Maybe it’s not that simple
either.
Gary
-----Original Message-----
From: Rich Cooper
[mailto:richcooper@mindspring.com]
Sent: Thu 3/24/2005 11:25 AM
To: Gary Berg-Cross; John F. Sowa
Cc: Jay Halcomb;
Alexander Povolotsky; abdoul@CYTANET.COM.CY;
standard-upper-ontology@LISTSERV.IEEE.ORG
Subject: Re: nature
-> "human brain" -> "language terms" ==>> knowledge
?
Gary Berg-Cross
wrote:
<snip/>
>Other parts of my mental
model lead to the idea of subjective judgments. Both are
mediated by models but the objective ones >are more like validated theory
and the subjective ones more like hypotheses. Or one
could speak of some not as tuned as
>others. Certainly it is not as easy to know whether
or not our "model" of peoples intentions "corresponds" or "agrees" with
the >person’s (assumed) real intentions – these are
more like pragmatic hunches and could be very affected by biases. We may
call >these types subjective, but there may be degrees and so the simple
dichotomy of objective to subjective is not the best.
Nice description, but
how do you represent degrees of subjectivity?
How would you define a gradient - number of hypotheses divided by
(hypotheses plus "validated" facts)?
One problem is that
the "validated" hypotheses of some people are what I would call subjective
errors. And they would think the same of my "validated"
hypotheses. So even the validation process is very subjective (or very
thirdish). So most people just treat it like pornography - they know
it when they see it.
Rich
|