AIList-REQUEST@SRI-AI.ARPA (AIList Moderator Kenneth Laws) (12/20/85)
AIList Digest Friday, 20 Dec 1985 Volume 3 : Issue 191
Today's Topics:
Queries - LISP Tools for the MicroVax II & Security Applications &
Intensional and Higher-Order Logic & Experience with the TI Explorer,
AI Tools - TI Superset of Common LISP & Object-Oriented Programming,
Psychology - Dreams & Lateral Thinking,
AI Tools - Equational Logic Programming Language
----------------------------------------------------------------------
Date: Thu, 19 Dec 85 11:41:14 GMT
From: Topexprs%cs.ucl.ac.uk@cs.ucl.ac.uk
Subject: LISP Tools and Environments for the MicroVax II
I'd like to know about any LISP based, KBS development environments for the
Microvax II. I'm looking for something that runs under microVMS, is up and
running OK, more or less debugged, reasonably documented, and available now;
is this too unrealistic? All useful pointers would be appreciated, also
approx. prices etc.
I'm used to LOOPS etc., so I'm not too keen on going to far down market.
Please reply to
HWB1.ARE @ CAM.PHX @ UCL-CS.ARPA
Thanks, Hal Blackburn.
------------------------------
Date: Thu, 19 Dec 85 07:43 CST
From: Sankar Virdhagriswaran <Araman@HI-MULTICS.ARPA>
Reply-to: Sankar Virdhagriswaran <Araman@HI-MULTICS.ARPA>
Subject: AI technology application to computer security
Is anyone out there aware of applications of AI technology for computer
security issues. I am aware of the pseudo object oriented models being
used in developing completely secure systems. Is there more of this
kind of research.
I will collect the responses and post it in the net
thanks in advance
send replies to Araman -at hi-multics
------------------------------
Date: 19 Dec 85 10:25:00 EDT
From: "CUGINI, JOHN" <cugini@nbs-vms.ARPA>
Reply-to: "CUGINI, JOHN" <cugini@nbs-vms.ARPA>
Subject: intensional and higher-order logic
Anyone doing any work on using intensional or higher-order logic
in knowledge representation? Any references (books, articles)
would be appreciated - thanks.
John Cugini <Cugini@NBS-VMS>
------------------------------
Date: Thu, 19 Dec 85 08:50:48 pst
From: Rolf Pfeifer
<pfeifer%ifi.unizh.chunet%ubc.csnet@CSNET-RELAY.ARPA>
Subject: Experience with the TI Explorer
How does the TI Explorer compare with other LISP Machines (e.g. Symbolics,
LMI, or XEROX)? In particular:
- availability of software
- compatibility with Symbolics (effort to transport programs developed on
Symbolics (or LMI) to the Explorer)
- subjective experiences after having used it (heavily) for some time
- support environment for AI applications development
- getting to know the Explorer (e.g. how useful are their utilities
they have apparently designed to support a new user?)
- performance
- other
Comments of any sort welcome.
Thanks.
--Rolf Pfeifer, University of Zurich, Switzerland
cernvax!unizh!pfeifer
------------------------------
Date: Tue 17 Dec 85 16:14:14-PST
From: Christopher Schmidt <SCHMIDT@SUMEX-AIM.ARPA>
Subject: Re: Superset of Common LISP...
[Forwarded from the Stanford bboard by Laws@SRI-AI.]
What TI's developing is a set of 4 4"*4" boards that implement
an MIT CADR including a number of chips almost at wafer scale. The
design differs from general purpose hardware only in the same way that
a CADR does and doesn't really implement "Common Lisp" per se.
When they say a "superset of Common Lisp" they mean MIT Lisp
Machine Lisp plus an additional package that implements Common Lisp.
This is rather like saying that a Shopsmith with the belt-sander
attachment is a $1500 belt-sander that happens to come with a free
Shopsmith. Whatever.
The processor is called a "Hummingbird" and TI gives a talk on
it locally every six months or so though they seem to have been by invitation.
--Christopher
------------------------------
Date: Wed 18 Dec 85 20:51:44-EST
From: Randy Haskins <rh%MIT-EECS@mit-eddie.MIT.EDU>
Subject: Object-Oriented Programming
Re: AIList Digest V3 #189
> Flavors: (send window-1 :expose)
> CORBIT: (expose window-1)
>
> What is seen as a message in Flavors is a function in CORBIT. So it is
> not the object, but the message which is functional. [...]
>
> Koenraad De Smedt, DESMEDT@HNYKUN52 (bitnet)
I have a hard time seeing the advantage to such a system, but then, I
was brought up on object-oriented programming. The advantages of
smart-objects/ dumb-messages is that functions that you write can be
more generic. I have written a fair number of utilities in ZetaLisp
for both ZL machines, and it's unbelievably quick and easy. One of the
functions I use a lot is called GET-OBJECTS-NAMED which follows.
(DEFUN GET-OBJECTS-NAMED (LST STRING &OPTIONAL (MSG :NAME))
(DELETE NIL
(MAPCAR
#'(LAMBDA (OBJ)(IF (STRING-SEARCH STRING (SEND OBJ MSG)) OBJ))
LST)))
(The actual function I wrote is much more general, but this is the part
that deals with Flavor instances pretty well.)
Since :NAME is a fairly common operation for most objects to support
(processes, most streams, zmacs-buffers), this function writes itself.
Another example is the fact that the PRINT function will see if its
argument is willing to handle a :PRINT-SELF method and will do that
instead of trying to deal with the object. The same is true for DESCRIBE;
I could go on for hours, but this article's already long enough.
Besides, I have to get back to my ZL-style Flavors I'm implementing in
MacLisp.
Random
------------------------------
Date: Wed, 18 Dec 85 15:55 CDT
From: Randy_Boys <boys%ti-eg.csnet@CSNET-RELAY.ARPA>
Subject: Response to query regarding dreams, the unconscious, and revelations
In response to a posting of 29-Nov-85 (G. Joly - Subconscious Reasoning:
Discovery and Invention)
My background is in physiological psychology and I was a staff member
at the University of Texas Health Science Center at Dallas' Cognitive
Psychophysiology Laboratory for five years, doing sleep research most of the
time. I have been working in the field of AI for three years now, always
trying to find avenues to blend my knowledge of cognitive and computer
science. I'm not sure that the following is very helpful to the progress of
either field, but I was compelled to respond to Gordon and Ken's comments
regarding "exceptional" insights that seemed to be associated with dream
state cognition.
Subconscious reasoning (if there is such a thing) is NOT the same as dream
state cognition. The cortex is active during REM sleep (I'm not going to go
into the REM = dream state issue) but may not be "processing" anything more
than random phasic events (similarly, I'm not going into the nature or
genesis of the dream state). Dreams do, however, exist and often reflect
"subconscious" elements. To assume that there is some particular significance
to this "window on the subconscious," as opposed to anything that may exist
for "conscious" cognition, would be stretching the state of our knowledge
(not that this should slow down the theorists!). In light of the fact that 2
billion people dream 4-5 times every night of their life and that only a few
(and usually highly personal) dream "revelations" are cited, my response to
postulations about dreams and discovery is "so what?" I do not believe that
this is an area of cognition that is well enough understood to direct our
inquiries into AI.
Randy Boys
boys@ti-eg
p.s. - after 30 years of heavily funded research, science can tell us what
sleep is not, but not what it is. Yes, you guessed it...the same is true of
dreaming.
------------------------------
Date: Thu, 19 Dec 85 12:12:43 GMT
From: gcj%qmc-ori.uucp@cs.ucl.ac.uk
Subject: 42
The term ``lateral thinking'' has been in the English language for
a some time. de Bono opposes lateral thinking to vertical thinking
in the following sense. Vertical thinking is the logical straight-
forward part of the mind, which is responsible for pattern
recognition, concerned with addition and gradual modification.
Lateral thinking is more concerned with making the best possible
use of the information that is already available, rearranging it
so that it is snapped out of the established pattern.
This is a paraphrase of de Bono's words, and it would take much
more space to give a really good account. The proposal that I want
to suggest is that de Bono's model fits into a plan for the hopes
of AI research. The vertical part of thought processes could be
carried out by a machine but the lateral (creative, humorous) part
could not.
In mathematics, the notion exists of ``factoring out'' onto a sub-
space, ie projecting down into a lower dimensional region. Perhaps
a defintion of machine intelligence could be that part of the mind
this left after the process of factoring out lateral subspace.
Gordon Joly,
gcj%qmc-ori@ucl-cs.arpa
------------------------------
Date: Fri, 20 Dec 85 9:14:41 EST
From: Robert Strandh <strandh@hopkins-eecs-bravo.ARPA>
Subject: Equational Logic Programming Language
EQUATIONAL LOGIC AS A PROGRAMMING LANGUAGE
Michael J. O'Donnell Robert I. Strandh
The University of Chicago The Johns Hopkins University
Department of Computer Science Department of Electrical Engineering
Ryerson Hall and Computer Science
1100 East 58th Street Barton Hall
Chicago, Illinois 60637 Baltimore, Maryland 21218
odonnell@uchicago.csnet strandh@hopkins.arpa
A processor for an Equational Logic Programming Language is available
for distribution to Berkeley UNIX 4.2BSD VAX installations. To get a
general idea of the capabilities of the interpreter, see "Programming
With Equations", by Christoph M. Hoffmann and Michael J. O'Donnell,
ACM ToPLAS, v. 4, no. 1 (January 1982) pp. 83-112. A user's manual is
included in "Equational Logic as a Programming Language", by Michael
J. O'Donnell, MIT Press, Cambridge, Massachusetts (1985).
The main novelties of the interpreter are 1) strict adherence to
simple semantics based on logical consequences of the given equations;
2) "lazy evaluation" (outermost evaluation) applied to all operators;
3) an implementation based on table-driven pattern matching, so that
there is no run-time penalty for large sets of equations.
A first experimental version was distributed in 1983. Recently,
Robert Strandh has replaced the table-driven interpreter written in
Pascal by compiled VAX machine code, producing an order-of-magnitude
improvement in performance. Preliminary timings indicate a
performance between interpreted and compiled Franz LISP.
The entire interpreter system, including source files, occupies about
3.5 megabytes. ARPANET users may acquire a copy on line by executing
ftp hopkins
then logging on as "anonymous'' with password "anonymous",
and executing the ftp commands:
cd pub
get equations.tar
The copying process will take an hour or more. After copying the
interpreter, please send a message indicating where you have installed
it. Others may acquire a 1600 BPI tape copy, in "tar" format, by
writing to Michael J. O'Donnell. Once the distribution file has been
acquired, it should be processed by the command
tar x (for the tape)
tar xf equations.tar (for the tar file)
Then, the instructions 2-4 in the file README should be followed. The
system is in the public domain, and may be copied freely. We request
notification from each site installing a copy. To defray the costs of
distribution, we also request a donation of $50 to The University of
Chicago for each tape, and of $5 to The Johns Hopkins University for
copies taken on the ARPANet.
------------------------------
End of AIList Digest
********************