[topicmapmail] RE: knowledge representation: some issues

[Paul Prueitt]

The "Prueitt" voting procedure has been implemented several times
algorithmically, and Yes I know of Pawlak's wonderful work and have visited
with Professor Zadeh about how the nature of Rough sets is quite different
from that of Fuzzy set theory. Zadeh and I agree that fuzzy set theory is a
means to get out of a category error that involves that assumption that
there is only one reality.  Rough sets (categories) are appropriate to
thinking about emergence and the issues of stratification of reality into
process layers.

Mental interpretation of the meaning of topics arranged in various fashion
is a stratified process in nature.  The formalism to shift scope and thus
the technical means for shifting meaning when XTM is used in practical
situations should (in my opinion) be stratified also.

Poland is recognized as having been the fatherland of yet another great
logician.

The voting procedure (that I developed) uses the notion of a rough category
policy.  The categories in the policy becomes crisp as a selection is made
via the voting of substructural elements.  These substructural elements are
such that they have come (somehow - perhaps algorithm, perhaps not) to
represent the item to be placed into a category.  In my model of human
memory these substructural elements are related to the memory of invariance
of textures, and colors and other archetypal things - not the remembrance -
but the substructure from which the remembrance arises.

http://www.bcngroup.org/area3/pprueitt/kmbook/Chapter5.htm

The use for this construct in the assembly of topic maps can be drawn fairly
easily from the four page description I have published at:

http://www.bcngroup.org/area3/pprueitt/kmbook/Appendix.htm

I wish to examine whether or not there is something of importance in this
work for the XTM process model.  I do not presume yet that there is, or that
the XTM community can or will embrace something that is not specified as
yet.

The two dimensional matrix (used for accounting purposes in the Prueitt
voting procedure) is to be used over and over at each of the decision points
that are taken to construct the relationships between topics, the properties
that are to be considered properly scoped within the set of topics and the
relationships (associations) between these.  The matrix is merely a means to
record the process of a rough process become crisp - and thus producing a
judgment.  You mistake its nature deeply and due to an absence of proper
consideration of what it is.

I am still having some difficulty following your messages.  I do still feel
that there is a disservice in spinning how you take authority on various
issues.  I ask you to be more careful.

You clearly have some interest in working through the spirit of the work
that I have done regarding Russian quasi axiomatic theory.  But you simply
must make a proper representation of my work.

As I have said many times lately - it is not about being polite anymore.

Once categories and the elements that are expected to coincide (in some
fashion - that may vary) are denoted, then there can be an openness of human
perceptual acuity (why not?) and external process perturbation of the voting
process.  However, the algorithms may be completely automated (even
autonomous).

The voting, like the voting for a national leader, is truly two level
(stratified) in that the vote of a single individual is almost always not
important.  It is the overall process that sorts out the entanglements and
leads eventually to a decision.

Stratified theory is attacked in various fashions often by individuals who
do not even themselves know why they are doing this.  It is the nature of
the immune system that protects the false perceptions about the nature of
our world.

Perhaps this framework is to ultimately to be useful to the discussions
about a process model for topic maps.





-----Original Message-----
From: W.M. Jaworski [mailto:wmj@gen-strategies.com]
Sent: Saturday, May 05, 2001 8:05 PM
To: Paul Prueitt; Robert Shaw; topicmapmail@infoloom.com
Cc: I.Witten@cs.waikato.ac.nz; ziarko@cs.uregina.ca
Subject: RE: knowledge representation: some issues


[Paul Prueitt]
I am just having difficult following what you are saying and why.

[wmj]
Is this a justification for using such strong words like "disservice"?

And now "ad rem":

[Paul Prueitt]
the voting procedure that I have developed might be uniquely useful in
the context of the process model for topic maps?  I looked at the list of
algorithms - these seem to be standard algorithms implemented in Java.

[wmj]
Is anything wrong with developing and using "standard algorithms"? Prof.
Witten might give you his justification.

Steps of my learning process:
(1) your "voting procedure" = an algorithm (aka process | procedure)
(2) process (algorithm) is executable by computer if represented by a
computer program
(3) your "voting procedure" is not programmable, therefore may assumption in
(1) is wrong

Conclusion: Human judgment is mandatory in your "voting procedure".

[Paul Prueitt]
I can
not assume from anything you have written that you personal know how this
formalism works or how it might be related to topic map shafts in scope.

[wmj]
You might be right on first point and you ARE RIGHT on the second.
It seems that it is your job to demonstrate the relevance. For example:
how nD topic maps can "benefit" from your 2D matrices at
http://www.bcngroup.org/area3/pprueitt/kmbook/Appendix.htm.?

In one of the submissions you mentioned fuzzy sets by Zadeh. More general
are rough sets by Z. Pawlak
http://www.cs.utep.edu/interval-comp/pawlak.html.

Thanks for introducing new tread. Is it going to be more productive? The
future will show. I will retire to other treads.

Best

WMJ


-----Original Message-----
From: Paul Prueitt [mailto:bcngroup@erols.com]
Sent: Saturday, May 05, 2001 5:30 PM
To: W.M. Jaworski; Robert Shaw; topicmapmail@infoloom.com
Cc: I.Witten@cs.waikato.ac.nz
Subject: knowledge representation: some issues


WMJ.

I am just having difficult following what you are saying and why.

You said:

"How "the formalism" is different from the formalism presented in the book
"Data Mining: Practical Machine Learning Tools with Java Implementations" by
Jan H. Witten, Eibe Frank - ISBN 1-55860-552-5? Public software at
http://www.cs.waikato.ac.nz/~ml/weka/ demonstrates complexity and
practicality of their formalism."

the formalism is the voting procedure for the tri-level architecture. I can
not assume from anything you have written that you personal know how this
formalism works or how it might be related to topic map shafts in scope.

Why do you cite this collections of algorithms in response to my suggestion
that the voting procedure that I have developed might be uniquely useful in
the context of the process model for topic maps?  I looked at the lsit of
algorithms - these seem to be standard algorithms implemented in Java.
Nothing in this list has even a significant percentage of the features of
the voting procedure - or the Russian quasi axiomatic theory - of which the
voting proceedure is my attempt to simplify. !  It is just a different type
of thing.

What is the point of the comparison.  I am confused.

Should we treat all formalism as the same value?

The scope might be elected by the voting procedure while at the same time
the actual topics with associations and properties might be gathered
together using the Mill's logic in

http://www.bcngroup.org/area3/pprueitt/kmbook/Chapter9.htm

However, if we talk about standards, then perhaps the relevance of your
communication is that there are many algorithms that can be used to process
the strengthen and inclusion of relationship and properties of topics.

Is this what you intent from your communcations to this forum?

Regarding your perceptions on knowledge management practice, to you see how
topic maps might be useful in knowledge management?