Copy Link
Add to Bookmark
Report
AIList Digest Volume 5 Issue 204
AIList Digest Friday, 28 Aug 1987 Volume 5 : Issue 204
Today's Topics:
Queries - GoldWorks & "If-Then" Rules & Conceptual Graphs &
Concept Definitions for Object Classification &
Modeling Creativity & How to Measure Learning Ability?,
Logic - Mr. S and Mr. P,
Review - Understanding AI,
Binding - Applied AI Reporter
----------------------------------------------------------------------
Date: Tue, 25 Aug 87 09:50 EDT
From: TAM%MCOIARC.BITNET@wiscvm.wisc.edu
Subject: GoldWorks by Gold Hill Computers
We have recently purchased GoldWorks expert system development
package, and have just returned from Gold Hill's 5-day training course.
I was wondering if anyone out there used GoldWorks and what is there
opinion of it compared to other PC-based expert system shells. I am
particularly interested in real applications using GoldWorks.
Thanks,
Paul Tam
Medical College of Ohio
bitnet: TAM@MCOIARC
------------------------------
Date: 24 Aug 87 16:38:52 +1000 (Mon)
From: "ERIC Y.H. TSUI" <munnari!aragorn.oz!eric@uunet.UU.NET>
Subject: Request for a set of "if-then" rules
I would like to have access to a set (or sets) of in-use "if-then" rules.
I am encoding a General Purpose Inference Engine (GPIE) for a conceptual
graph (J.F. Sowa's book on Conceptual Structures) system and would like to
ask for a set of "if-then" rules currently used in some typical rule-based
systems. Something between 50 to 200 rules are quite appropriate for testing.
These rules, presumably not written in the CG notation, will be encoded in
the CG formalism and then used for testing the GPIE. The rules,
encoded in the CG notation, can be returned to the original source.
Meta-rules, self-referencing rules and other non base level rules are also
sought, though they may not incorporated into the testbed.
Rules are most preferred in areas like financial and auditing domains,
diagnostic systems, planning and intelligent help/advisory systems.
Others are also welcome.
Does anyone have access to such a set of rules ?
Can someone provide pointers to locate such rules ?
Any other advise ?
Eric Tsui eric@aragorn.oz
Division of Computing and Mathematics
Deakin University
Victoria 3217 AUSTRALIA
------------------------------
Date: 25 Aug 87 10:26:00 EST
From: cugini@icst-ecf.arpa
Reply-to: <cugini@icst-ecf.arpa>
Subject: conceptual graphs
>
> Date: 21 August 1987, 08:42:59 EDT
> From: Anand Rao <ANAND@ibm.com>
> Subject: Conference - Conceptual Graphs
>
>
> SECOND ANNUAL WORKSHOP ON CONCEPTUAL GRAPHS
>
> IBM Paris Scientific Center
> 2, 3, 4 September 1987
>
> ....
Regarding this recent announcement on AIlist: Are there any articles
currently in print on this topic? Eg, Proceedings of the 1st
conference? I'd be very grateful for any pointers you can give me on
this. Tutorial/basic-overview type articles are especially welcome.
John Cugini <Cugini@icst-ecf.arpa>
------------------------------
Date: 27 Aug 87 12:20:08 +1000 (Thu)
From: "BETTY CHENG" <munnari!aragorn.oz!cheng@uunet.UU.NET>
Subject: Request for a set of Concept Definitions for Object
Classification
I am doing a project on Concept Classification and would like to know
whether anyone has developed, have access to or can provide pointers to
locate a set of concept definitions for real world knowledge.
For example, definitions for physical objects like bus, vehicle,
clothes etc. Definitions for primitive acts are also
sought. For example, definiton for 'open', 'close', 'eat' etc.
These definitions of concepts will be encoded in the conceptual graph
(c.f. Sowa's book on Conceptual Structures) formalism and an interactive
classification program, currently being implemented, will help to
identify instances of existing concepts, insert new concepts into the
knowledge base and in the process of the executing the above two functions,
perform exact and partial matching on existing concepts.
Ideally, these definitions are already presented in a frame-based or
in an attribute-value pair form. However, other forms of presentations will
certainly be considered.
Could anyone possess such a set of definitions of concepts ?
Any idea on where I can locate a set of these definitions ?
Any comment/criticism/idea ?
[A summary of the result of the responses will be posted, given that
there exists sufficient interest in the readership.]
Betty Cheng cheng@aragorn.oz
------------------------------
Date: 27 Aug 87 03:35:52 GMT
From: harnad@princeton.edu
Subject: Modeling Creativity
I would be grateful to receive references to work on modeling
creativity (in any domain -- verbal, mathematical, artistic, motor). I
am also interested in relevant experimental and observational work.
Stevan Harnad harnad@mind.princeton.edu (609)-921-7771
--
Stevan Harnad harnad@mind.princeton.edu (609)-921-7771
------------------------------
Date: 27 Aug 87 20:05:40 GMT
From: berke@locus.ucla.edu
Subject: How to measure learning ability?
I am involved with a project part of which is to teach mime to
learning-disabled children. I maintain that: Mimicry behavior
is integral to or forms the basis for animal learning. Directly
training mimicry should therefore directly train learning
ability.
This is a simple, to me, an obvious claim. The question is, how
to determine whether it is true? If you can answer the following
simple question, I would appreciate hearing from you:
Question 1: Are there any measures of general learning ability
that are commonly accepted? If not, what measures of learning
ability do you use (or know of), whether they purport to measure
verbal learning, skill acquisition, or any other behavior that
can be classified as learning?
In "cognitive science" and related fields there is a lot of hubub
currently about new and better brain models. I have my own which
I call Network Recombination. I refer to "learning and memory" as
a unified process of learning/memory because of the implications
of my model. If you ascribe to a model of how brain activity
produces the phenomenon of learning/memory (or "learning" or
"memory" separately), I would appreciate an answer to the
following question:
Question 2: Does your model make any predictions about
intermodal transfer of abilities? Specifically, say a subject's
verbal skills are poor and so she does poorly on vocabulary tests
which Thorndike (in Human Learning, 1931, p.174) considers "an
excellent intelligence test." Say I now train the subject in
physical skills to increase discrimination, analysis, and
creative abilities (defined primitively below). How much will
these abilities transfer to verbal or quantitative skills? What
form will the transfer take? Or will there be none?
Discrimination ability - seeing different parts in observation
Analytical ability - breaking things into parts
Creative ability - putting parts into new wholes
I appreciate all opinions and advice, especially from people who
have worked with learning-disabled children (and even more from
people to whom this posting seems based on my ignorance and
misconceptions). But I would like to specifically request those
putting forth "mind models" for predictions of their models. I
would like to QUANTIFY increase in general learning ability, and
so the predictions of specific models with specific properties is
necessary.
If there is no such thing as general learning ability, then I
would like to QUANTIFY transfer of abilities from physical to
verbal or intellectual skills, or verify that there is none.
Can you help me?
I have posted this to several news groups. Perhaps it would be
best to reply to me in e-mail, or to decide on a single group for
follow-ups, perhaps sci.research. It has little traffic.
Thank you in advance for all replies.
Peter Berke
berke@cs.ucla.edu
(213) 394 - 6797
------------------------------
Date: Mon, 24 Aug 87 13:37:25 EDT
From: tim@linc.cis.upenn.edu (Tim Finin)
Subject: Mr. S and Mr. P
I don't know the full history of the Mr. S and Mr. P puzzle, but I'm
sure it goes back a long way. I first saw it when Will Dowling of
Drexel University sent it to me with the suggestion that it might be
easy to express the solution in Prolog. I posted it to Prolog-digest
in November or October of 1984. There were several follow up posting,
some of which discussed its history. You could explore the prolog
digest archives at SUMEX-AIM for more information.
Here is the puzzle as it was told to me:
There are 2 integers n and m between 3 and 98 inclusive. Mr. S has
been told their sum and Mr. P their product. The following truthful
conversation occurs:
P: I don't know n and m.
S: I knew you didn't. Neither do I.
P: Now I know them!
S: Now I do, too!!
What are the values of n and m?
By the way, the answer to this version of the puzzle is NOT 4 and 13!
I won't spoil anyone's fun by saying what the answer is, or by
describing how one determines what it is or is not.
This is a simple example of a general problem which requires one to
model and reason about the beliefs and knowledge of other agents, a
topic that has been receiving some attention lately. Halperin (IBM)
and Moses (Stanford) have been using a similar puzzle (variously
called "the cheating wives" and "the dirty children") in their recent
work.
Tim
------------------------------
Date: 23 Aug 87 20:41:00 GMT
From: uxc.cso.uiuc.edu!osiris!goldfain@a.cs.uiuc.edu
Subject: Re: A good language-independent text on
To: murray@cs.strath.ac.uk
Subject: Good AI Survey Book
The book: Understanding Artificial Intelligence, by: Henry C. Mishkoff,
published by Howard W. Sams & Company, a division of MacMillan Inc. (4300
West 62nd Street, Indianapolis, IN 46268 USA) seems to have about the emphasis
and level that you describe. You may want to check into it. Here in the
U.S., I got it for about $15.
It was written for industry audiences more than university students, but you
may consider this a plus, judging from your "application" emphasis. In
addition to a high-quality typesetting, it has a bibliography and glossary in
the back. It's 250 pages. (NOTE: I have no vested interest in whether or not
the author sells a single copy.)
- Mark Goldfain
(ARPA: goldfain@osiris.cso.uiuc.edu)
------------------------------
Date: 24 Aug 87 18:51:13 GMT
From: gatech!pitt!psuvax1!cisunx!wvucsb.UUCP!aw@seismo.CSS.GOV (Ajay
Waghray)
Subject: Re: Applied AI Reporter??
> Approved: ailist@stripe.sri.com
> Xref: wvucsb comp.ai.digest:300
>
> I have heard of a publication called the Applied AI Reporter, from the
> University of Miami. Can someone give me an address or phone number for
> more information on the publication?
>
> Thanks,
> Larry
> -------
The Applied AI Reporter published from the Intelligent Computer
Systems Research Institute, University of Miami.
I do not know any phone numbers but the address for editorial correspondance
is ::
Editor
P.O. Box 248235
Coral Gables
FL 33124
Address for subscriptions and inquiries is ::
ICS RESEARCH INSTITUTE
P.O. Box 1308-EP, Fort Lee
NJ 07024
Hope this helps
Ajay
----
{allegra, cadre, bellcore, psuvax1}!pitt!wvucsb!aw
------------------------------
Date: 25 Aug 87 17:17:22 GMT
From: root@sgi.sgi.com (Superuser)
Subject: Re: Applied AI Reporter??
It is published by the Intelligent Computer Systems Research
Institute, 421C Jenkins Bldg., Stanford Drive, Coral Gables,
FL 33124 (305) 284-5195 BITNET dumics@ser
Mike Bender
sgi!wdl1!mhb
mhb@wdl1.UUCP
------------------------------
Date: Wed, 26 Aug 87 09:47:12 PDT
From: lambert@cod.nosc.mil
Subject: applied ai reporter
Larry,
In response to your request in AIList:
Applied Artificial Intelligence Reporter
Intelligent Computer Ssytems (ICS) Research Institute
University of Miami
P.O. Box 248235
Coral Gables, FL 33124
$98.00 per year.
Dave
------------------------------
End of AIList Digest
********************
