AIList-REQUEST@SRI-AI.ARPA (AIList Moderator Kenneth Laws) (09/03/85)
AIList Digest Monday, 2 Sep 1985 Volume 3 : Issue 116
Today's Topics:
Queries - Parallel LISP Environments & Hypercube Simulators &
OPS5 in Interlisp & Texture Images & Tax Expertise,
Philosophy - Intelligence and Generalization,
Knowledge Representation - Limitations of Frames/Semantic Nets,
Linguistics - CSLI Reports,
Expert Systems - Psychiatry
----------------------------------------------------------------------
Date: Tue, 27 Aug 85 15:37:18 EDT
From: Deepak Kumar <kumard%buffalo.csnet@csnet-relay.arpa>
Subject: Parallel LISP Environments
We are planning to implement a parallel-environment
simulator in LISP for trying out various control
metaphors that require parallelism.
Anyone already having any implementations or
experiences in using such environments?
We would appreciate any kind of responses that could
help in highlighting various aspects of design as well
as implementation.
Thanx.
Deepak.
UUCP : {cmcl2,hao,harpo}!seismo!rochester!rocksvax!sunybcs!kumard
...{allegra,decvax,watmath}!sunybcs!kumard
CSNET : kumard@buffalo
ARPA : kumard%buffalo@csnet-relay
BITNET : kumard@sunybcs
------------------------------
Date: 30 Aug 85 13:54:23 EDT (Fri)
From: duke@mitre.ARPA
Subject: Simulators for hypercube architectures
Lisa Sokol (sokol@mitre) and I are investigating the use of parallel
architectures for object-oriented simulations. We have a large simulation
called the BEM (Battlefield Environment Model) which is written in the
simulation language ROSS, which runs on top of Franz LISP on a VAX. We
would like to have a hypercube simulator which would help us investigate
strategies for distributing objects in a hypercube. Our main problem seems
to be the integration of LISP code with the simulators. Translating the
Lisp into C would be messy at best, even with a translator program. Another
possibility would be to have Franz LISP running in another process on the
VAX and communicating with the simulator process. That sounds pretty messy
also, particularly for analyzing the timing since the Lisp is running
outside of the simulator process. For our purposes, it may be better to
write a Lisp simulator of a hypercube architecture. Do you know of anyone
who already has such a simulator? We expect that Lisp will be available on
some hypercube in the next year, and our simulator studies should allow us
to determine how to distribute the objects by that time.
Duke Briscoe
duke@mitre
------------------------------
Date: Mon, 2 Sep 85 00:20:55 cdt
From: neves@wisc-ai.arpa (David Neves)
Subject: OPS5 in Interlisp?
Has anyone translated OPS5 into Interlisp? If yes, please send me
mail.
David Neves
Computer Sciences Department
University of Wisconsin-Madison
Usenet: {allegra,heurikon,ihnp4,seismo,uwm-evax}!uwvax!neves
Arpanet: neves@uwvax
------------------------------
Date: Thu 29 Aug 85 14:48:30-PDT
From: BARNARD@SRI-AI.ARPA
Subject: texture images
I am collecting a set of digital images of natural textures. If you
have any such images, I would appreciate your help. They could either
be images of single textures (Brodatz-type) or images of natural
scenes constaining several textured surfaces. I will make the
collection available to the AI community.
Reply to Barnard@SRI-AI.
------------------------------
Date: Thu 29 Aug 85 13:26:29-CDT
From: Charles Petrie <CS.PETRIE@UTEXAS-20.ARPA>
Subject: Tax Expertise Available for Expert System
Prof. Lewis Solomon is a specialist in tax law and is interested in
working with someone on an expert system in that domain. He would also
like to hear about existing systems. His US Mail address is:
George Washington University
National Law Center
Washington, D.C. 20052 Phone #(202)676-6753
------------------------------
Date: 28 Aug 85 14:51:00 EDT
From: "CUGINI, JOHN" <cugini@nbs-vms>
Reply-to: "CUGINI, JOHN" <cugini@nbs-vms>
Subject: Intelligence as the ability to apply the distributive property
To wax philosophical for a while and then get down to specifics:
1. A major (the major?) aspect of intelligence is the ability to
see several instances of a thing and then be able to abstract
out common features. This is terribly handy for information
representation as it allows one to condense, by "factoring out",
the total info. Nothing really new here - the notion of
"inheritance", based on the is-a relation is a clear example.
2. The whole problem of context seems strongly related - isn't
context simply that which we hold constant over time, until we
are forced to assume something new? - finding the antecedent
of pronouns can be seen as the problem of: given the most
recent n sentences, what (single, common) interpretation allows
me to maximize the number of unchanged assumptions?
3. A big issue then becomes, how rich is the repertoire of
pattern-matching mechanisms? Put concretely, and blithely
assuming that you can represent anything with lists, given n
lists (Lisp-type if you like), which abstractions are available
for describing (by generalizing about) them, and which of several
applicable abstractions should take precedence? For example,
is there a Lisp system out there which would abstract over the
following sets of lists the way I suspect would be done by thee
or me? Any research being done on general-purpose similarity
finding among Lisp lists?
(a b c), (a b c), (a b c)
(a b c), (a b w), (a b q), (a b d e r)
(a b c q), (a b c e), (a b c (d e (w e)))
(a b c), ((a) (b) (c)), (((a)) b c)
(a b c), (aaaa bb ccccc), (aaa b cccc)
(a b c), (A B C), (a B c)
((a b c) (b)), ((a b c d e f) (b c d e)), ((a b (c d) (e f)) (b (c d)))
(a a a), (b b b b b b), (xx xx xx xx)
(2 3 5), (5 6 11), (12 2 14)
(2 3 5), (5 6 11), (12 2 14), (one three four), (3 four 7)
The last set is, of course, somewhat unfair, since it relies on
knowledge of English semantics, and not simply on structural
properties.
John Cugini <Cugini@NBS-VMS>
Institute for Computer Sciences and Technology
National Bureau of Standards
Bldg 225 Room A-265
Gaithersburg, MD 20899
phone: (301) 921-2431
------------------------------
Date: 28 Aug 1985 10:04-EDT
From: Robert.Frederking@CMU-CS-CAD.ARPA
Subject: AI Mag article on KR
The current (Fall 85) issue of AI Magazine has an article by
Ron Brachman raising one of the issues (lack of definitional
information) that has been bothering me about AI knowledge bases. He
explains why these capabilities are necessary, and then shows that they
aren't possible in a KR system where prototype information can be
overridden, i.e., where nodes are treated as prototypes containing
default information.
[Among other points, Brachman shows that a frame/semantic net/
inheritance network/nonmonotonic representation seems to have the
following limitations: it cannot determine automatically where in
the network to place an ELEPHANT-WITH-THREE-LEGS (or similar composite)
node; it cannot infer that this is a subclass of elephants, or indeed
that it has anything to do with elephants; it cannot infer that instances
have three legs; and it cannot recognize a three-legged elephant as
an instance of this concept. My 5-year-old also enjoyed the riddles
about "What's big and gray, has a trunk, and lives in the trees?"
(Valid answers in Ron's strawman representation systems are "elephants",
"giraffes", and "ideas".) -- KIL]
------------------------------
Date: Wed 28 Aug 85 17:10:02-PDT
From: Emma Pease <Emma@SU-CSLI.ARPA>
Subject: New Reports
[Excerpted from the CSLI Newsletter by Laws@SRI-AI.]
NEW CSLI REPORTS
Report No. CSLI-85-29, ``Equations, Schemata and Situations: A
framework for linguistic semantics'' by Jens Erik Fenstad,
Per-Kristian Halvorsen, Tore Langholm, and Johan van Benthem, and
Report No. CSLI-85-30, ``Institutions: Abstract Model Theory for
Computer Science'' by J. A. Goguen and R. M. Burstall, have just been
published. These reports may be obtained by writing to David Brown,
CSLI, Ventura Hall, Stanford, CA 94305 or Brown@SU-CSLI.
------------------------------
Date: Thu, 29 Aug 85 15:49 EST
From: Clarke Thacher <UKC323%UKCC.BITNET@WISCVM.ARPA>
Subject: Expert Systems In Psychiatry (Responses)
I would like to thank all of the people who responded to my recent
request. The following is a summary of the responses which I have
received (so far). I have passed this information to Dr. David Atcher
in our Med. School, and he may be following up on some of these leads.
_______
> Date: 26 August 1985, 14:45:02 GMT
> From: Thomas Rothenfluh K715311 at CZHRZU1A
>
> I've read your request in the AIList and like to submit you two hints.
>
> In the last AI Magazine there was an article about AI research at
> the University of Michigan where some research is done in the field
> of expert systems in psychiatry. Upon request, I received a manuscript
> with a short overview about current systems, own work and bibliography:
> Micheal Feinberg & Robert K. Lindsay
> "Expert systems in psychiatry"
> (Mental Health Institute, Dept. of Psychiatry, Univ. of Michigan,
> Ann Arbor, MI, 48109-0010)
>
> Another research group around Janet L. Kolodner at the Georgia Institute
> of Technology (School of Information and Computer Science, Georgia
> 30332) is also working on a psychiatric expert system, e.g. see:
> Kolodner, J.L.: Towards an understanding of the role of experience
> in the evolution from novice to expert. In: M.J. Coombs. (Ed).
> Developments in expert systems. London: Academic, 1984.
>
> There are a lot of references in the field of computer-assisted
> diagnosis in psychiatry, however there a only few concerning
> expert systems in a sense that deserves that term.
>
> As I'm working on a dissertation project related to this field
> (a computer model for the diagnostic process of experts in diagnosing
> borderline personality disorders), I would be happy if you could send
> me a collection of other people's answers.
>
> Sincerely yours,
>
> Thomas E. Rothenfluh, research assistant
> Konfliktforschungsstelle
> University of Zurich
> Zurichbergstrasse 43
> CH-8044 Zurich
> Switzerland (K715311 at CZHRZU1A.BITNET)
_______
> Date: Mon 26 Aug 85 17:12:45-EDT
> From: David.Servan-Schreiber@CMU-CS-C.ARPA
>
> I have developed with a collegue an expert system for the
> pharmacological treatment of depressive disorders called Blue-Box. You
> can find a description of our work in "Computers in Biomedical
> Research", Feb 84 issue. and also in the KSL series published by
> Stanford University (public domain reports).
>
> David
_______
> Date: Tue 27 Aug 85 18:19:24-PDT
> From: Christopher Schmidt <SCHMIDT@SUMEX-AIM.ARPA>
>
> I don't know much about his work, but Sam Holtzman
> (Holtzman@sumex-aim.arpa) has an expert system which helps infertile
> couples decide what their next move should be. You could ask him for
> other references at least.
> --Christopher
_______
> Date: Wed, 28 Aug 85 10:14:03 pdt
> From: broder@isi-vaxa (Ben Broder)
>
> I would be interested in any leads you turn up on Psychiatric
> Expert Systems. I'm sorry that I don't have any information to give
> you.
>
> Ben Broder
_______
> Date: Thu 29 Aug 85 11:15:27-PDT
> From: Michael Walker <WALKER@SUMEX-AIM.ARPA>
>
> You asked on the AIList for pointers to expert systems in
> psychiatry. One developed here at Stanford by Benoit Mulsant and David
> Servan-Schreiber was written up in Computers and Biomedical Research,
> as I recall in 1984 or late 1983.
>
> Mike Walker
------------------------------
End of AIList Digest
********************