Copy Link
Add to Bookmark
Report
AIList Digest Volume 2 Issue 028
AIList Digest Saturday, 10 Mar 1984 Volume 2 : Issue 28
Today's Topics:
Games - SMAuG Player Simulation,
Mathematics - The Four-Color Theorem,
AI Tools - Interlisp Availability,
Review - Playboy AI Article,
Expert Systems - Computer Graphics & Hardware/Software Debugging,
Expert Systems - Production Tools,
Review - Laws of Form
----------------------------------------------------------------------
Date: 8 Mar 84 17:17:53 EST
From: GOLD@RU-BLUE.ARPA
Subject: a request for suggestions....
Some of you may be aware of the project known as SMAuG (Simultaneous
Multiple AdventUrer Game) that is ongoing at Rutgers University. It is
an applied research project designed to examine the problems of distrib-
uting the work of a complex piece of software accross local intelligent
devices and a remote timesharing computer. The software is a multiple
player adventure game.
Within the game a player may interact with other players, or with
software controlled players referred to as Non Player Characters (NPC's).
The NPC's are the area of the project which I am personally involved
with and for which I write to this bboard. There are many interesting
subtopics within the NPC issue. NPC communication, self mobility,
acquisition of knowledge, and rescriptability just to name a few.
The object is to create an NPC which can interact with a player character
without making it obvious that the it is machine controlled and not
another player character. [Aha! Another Turing test! -- KIL]
I would like to request suggestions of relevent publications that I
should be familiar with. This is a large project, but I am loathe
to make it even larger by ignoring past work that has been done.
I would greatly appreciate any suggestions for books, journal articles,
etc. that might offer a new insight into the problem.
Please send responses to Gold@RU-Blue.
Thank you very much,
Cynthia Gold
------------------------------
Date: Thu 8 Mar 84 10:54:46-PST
From: Wilkins <WILKINS@SRI-AI.ARPA>
Subject: Re: The Four-Color Theorem
I am not familiar with the literature on the 4-color proof, nor with whether it
is commonly accepted. I do however have a lot of experience with computer
programs and have seen a lot of subtle bugs that do not surface 'til long after
everyone is convinced the software works for all possible cases after having
used it. The fact that another person wrote a different program that got the
same results means little as the same subtle bugs are likely to be unforeseen
by other programmers. If the program is so complicated that you cannot prove
it or its results correct, then I think the mathematicians would be foolish
to accept its output as a proof.
David
------------------------------
Date: Thu 8 Mar 84 09:59:26-PST
From: Slava Prazdny <Prazdny at SRI-KL>
Subject: Re: The Four-Color Problem
re: the 4-color problem
A nice overview paper by the authors is in "Mathematics Today",
L.A.Steen (ed),Vintage Books, 1980.
------------------------------
Date: 8 Mar 1984 11:22-PST
From: Raymond Bates <RBATES at ISIB>
Subject: Interlisp Availability
A version of Interlisp is available from ISI that runs on the VAX
line of computers. We have versions for Berkeley UNIX 4.1 or 4.2
and a native VMS version. It is a full and compete
implementation of Interlisp. For more information send a message
to Interlisp@ISIB with your name and address or send mail to:
Information Science Institute
ISI-Interlisp Project
4676 Admiralty Way
Marina del Rey, CA 90292
Interlisp is a programming environment based on the lisp
programming language. Interlisp is in widespread use in the
Artificial Intelligence community. It has an extensive set of
user facilities, including syntax extensions, uniform error
handling, automatic error correction, an integrated
structure-based editor, a sophisticated debugger, a compiler and
a file system.
P.S. I just got AGE up and running under ISI-Interlisp (the new
name of Interlisp-VAX) and will start to work on EMYCIN soon.
/Ray
------------------------------
Date: Thu 8 Mar 84 20:35:02-CST
From: Werner Uhrig <CMP.WERNER@UTEXAS-20.ARPA>
Subject: Playboy 4/84 article: AI-article by Lee Gomes
If you needed an excuse to read playboy (even deduct it from your taxes ?? )
on page 126 is an article:
The Mind of a New Machine. can the science of artificial intelligence
produce a computer that's smarter than the men who build it?
nothing earth-shaking, a little history, a little present state of the art,
a little outlook into the future. but, it's interesting what's being fed
to this audience. Something to hand to a friend who wants to know what
this is all about, and doesn't mind getting side-tracked by "Playmates
Forever" on page 129.
Enjoy or Suffer, it's your choice.
------------------------------
Date: Thu 8 Mar 84 16:55:31-PST
From: Ken Laws <Laws@SRI-AI.ARPA>
Subject: Expert Systems in Computer Graphics
The February issue of IEEE Computer Graphics and Applications has a short
blurb on Dixon and Simmons' expert system for mechanical engineering design.
Following the blurb, on p. 61, is the notice
IEEE Computer Graphics and Applications is planning an issue
featuring articles on expert systems in computer graphics
applications in early 1985. Those interested in contributing
should contact Carl Machover, Machover Associates, Inc., 199
Main St., White Plains, NY 10601; (914) 949-3777.
The issue also contains an article on "Improved Visual Design for Graphics
Display" by Reilly and Roach. The authors mention the possibility of
developing an expert consulting system for visual design that could be
used to help programmers format displays. (I think automated layout
for the graphics industry would be even more useful, and an excellent
topic for expert systems research.) They cite
J. Roach, J.A. Pittman, S.S. Reilly, and J. Savarse, "A Visual
Design Consultant," Int'l Conf. Cybernetics and Society, Seattle,
Wash., Oct. 1982.
as a preliminary exploration of this idea.
-- Ken Laws
------------------------------
Date: Fri 9 Mar 84 17:23:30-PST
From: Ken Laws <Laws@SRI-AI.ARPA>
Subject: Expert System for Hardware/Software Debugging
The March issue of IEEE Computer has an article by Roger Hartley
of Kansas State University on the CRIB system for fault diagnosis.
The article starts with a discussion of expertise among experts
vs. that among practitioners, and about the process of building
a knowledge base. Hartley then introduces CRIB and discusses, at
a fairly high level, its application to fault diagnosis in ICL 2903
minicomputers. He then briefly mentions use of the same hierarchical
diagnostic strategy in debugging the VME/K operating system.
This article is an expanded version of the paper "How Expert Should an
Expert System Be?" in the 7th IJCAI, 1981.
-- Ken Laws
------------------------------
Date: 8 March 1984 1426-est
From: Roz <RTaylor.5581i27TK @ RADC-MULTICS>
Subject: Expert Systems Production tools
To all who have queried me regarding what info I have or have received on
expert systems production tools...I must apologize. Have not gotten it
into suitable format as yet; I am literally behind the power curve
with some new efforts (high visibility) recently assigned to me (approx
4 weeks ago--about the time I could start editing what I have). I will
post it to the AIList, but unless something helps it won't be before
April. Unfortuanately, what has already been massaged is in 132 char
[tabular] format and would not post easily to the list that way. I am
sorry, folks. But I have not forgotten you.
Roz
------------------------------
Date: 7 Mar 84 19:12:34 PST (Wed)
From: Carl Kaun <ckaun@aids-unix>
Subject: More Laws of Form
Before I say anything, you all should know that I consider myself at best
naive concerning formal logic. Having thus outhumbled myself relative to
anyone who might answer me and having laid a solid basis for my subsequent
fumbling around, I give you my comments about Laws of Form. I do so with the
hope that it stirs fruitful discussion.
First, as concerns notation. LoF uses a symbol called at one point a
"distinction" consisting of a horizontal bar above the scope of the
distinction, ending in a vertical bar. Since I can't reproduce that very
well here, I will use parentheses to designate scope where the scope is
otherwise ambiguous. Also, LoF uses a blank space which can be confusing. I
will use an underline "_" in its place. And LoF places symbols in an
abutting position to indicate disjunction. I will use a comma to separate
disjunctive terms.
In Lof, the string of symbols " (a)|, b ", or equivalently, " a|, b" is
equivalent logically to the statement " a implies b". The comparison with
the equivalent statement " (not a) or b" is also obvious. The "|" symbol
seems to be used as a postfix unary [negation] operator. "a" and "b" in the
formulae are either "_" or "_|" or any allowable combination of these in
terms of the constructions available through the finite application of the
symbols "|" and "_". LoF goes on to talk about this form and what it implies
at some length. Although it derives some interesting looking formulae (such
as the one for distribution), I could find nothing that cannot be equivalently
derived from Boolean Algebra.
Eventually, LoF comes around to the discussion of paradoxical forms, of which
the statement "this sentence is false" is the paradigm. As I follow the
discussion at this point, what one really wants is some new distinction (call
it "i") which satisfies the formula " (i|)|, i". At least I think it should
be a distinction, perhaps it should also be considered simply to be a symbol.
The above form purports to represent the sentence "this sentence is false".
The formulation in logic is similar to the way one arrives at complex
numbers, so LoF also refers to this distinction as being "imaginary". At
this point I am very excited, I think LoF is going to explore the formula,
create an algebra that one can use to determine paradoxical forms, etc. But
no development of an algebra occurs. I played around with this some years
ago trying to get a consistent algebra, but I didn't really get anywhere
(could well be because I don't know what I'm doing). Lof goes on to describe
the distinction "i" in terms of alternating sequences of distinctions,
supposedly linking the imaginary distinction to the complex number generator
exp(ix), however I find this discussion most unconvincing and unenlightening.
Now LoF returns to the subject of distinction again, describing distinctions
as circles in a plane (topologically deformable), where distinction occurs
when one crosses the boundary of a circle. In this description, the set of
distinctions one can make is firmly specified by the number of circles, and
the ways that circles can include other circles, etc. LoF gives a most
suggestively interesting example of how the topology of the surface might
affect the distinctions, and even states that different distinctions result
on spheres than on planes, and on toroids than on either, etc. Unfortunately
he does not expound in this direction either, and does not link it to his
"imaginary" form above, and I think I might have given up on LoF at this
time. LoF doesn't even discuss intersecting circles/distinctions.
The example that LoF gives is of a sphere where one distinction is the
equator, and where there are two additional distinctions (circles,
noninclusive one of the other) in the southern hemisphere. Then the
structure of the distinctions one can make depends on whether one is in the
northern hemisphere, or in the southern hemisphere external to the two
distinctions there, or inside one of the circles/distinctions in the southern
hemisphere. As I say, I really thought (indeed think today) that perhaps
there is some meat to be found in the approach, but I don't have the time to
pursue it.
I realize that I have mangled LoF pretty considerably in presenting my
summary/assessment/impressions of it. This is entirely in accordance with
my expertise established above. Still, this is about how much I got out of
LoF. I found some suggestive ideas, but nothing new that I (as a definite
non-logician) could work with. I would dearly love it if someone would show
me how much more there is. I suspect I am not alone in this.
Carl Kaun ( ckaun@AIDS-unix )
------------------------------
End of AIList Digest
********************
