Copy Link
Add to Bookmark
Report
AIList Digest Volume 2 Issue 058
AIList Digest Tuesday, 15 May 1984 Volume 2 : Issue 58
Today's Topics:
AI Tools - Personal Computer Query,
AI Books - LISPcraft Review,
Humor - Distributed Intelligence,
Linguistics - Metaphors,
Job Market - Noncompetition Clauses,
Seminar - Content-Addressable Memory,
Conference - IEEE Knowledge-Based Systems Conference
----------------------------------------------------------------------
Date: 11 May 84 13:13:39-PDT (Fri)
From: decvax!wivax!apollo!tarbox @ Ucb-Vax
Subject: LISP machines question
Article-I.D.: apollo.1f4a00cb.d4d
Can anyone out there tell me what the smallest,
(ie. least expensive) home/personal computer is
that run some sort of LISP?
-- Brian Tarbox @APOLLO
------------------------------
Date: Wed, 9 May 84 17:36:35 pdt
From: wilensky%ucbdali@Berkeley (Robert Wilensky)
Subject: AIList book announcement
I want to dispel an incorrect impression left by Dave Touretzky about my
recent book on LISP (which, incidentally, is called
LISPcraft
by Robert Wilensky
W. W. Norton & Co.
New York, 1984. Softcover, 385 pages, $19.95 list. )
Specifically, Touretzky gave the impression the my book was geared to
advanced Franz LISP programming, and was not appropriate as a general
tutorial for the novice. Nothing could be further from the truth.
LISPcraft is NOT meant to be primarily a reference for Franz LISP, nor is it
intended as an advanced LISP text. Rather, the book is meant to be a
self-contained LISP tutorial for the novice LISP programmer.
LISPcraft does assume some familiarity with computers, so it may not be
ideal for the computationally illiterate. On the other hand, like
Touretzky's book, and unlike Winston and Horn's, almost the entire length of
my book is a tutorial on various aspects of the language.
From my point of view, the primary difference between these books is that I
try to cover the language from the programmer's point of view. This means
that I pay homage to the way LISP programmers actually use the language. As
a consequence, I spend some time on features of LISP that one hardly finds
discussed anywhere, e. g., programming idioms, macro writing techniques,
read macros, debugging, error handling, non-standard flow of control, the
oblist, non-s-expression data types, systems functions, compilation, and
aspects of I/O. I also give some serious programming examples (pattern
matching and deductive data base management). However, my book starts at
ground zero, and works its way through the basics. In fact, the text is
about evenly divided between the sort of issues listed above and more basic
``car-cdr-cons'' level stuff. Most importantly, the text is entirely
tutorial in nature and presumes no previous knowledge of LISP whatsoever. I
believe that basics of LISP programming are presented to the uninitiated as
well here as they are anywhere.
In sum, LISPcraft contains a more extensive exposition of LISP than either
Winston's or Touretzky's book. Winston's book contains many more examples of
LISP programs than does LISPcraft, and Touretzky's book covers less material
at a slower pace.
As Touretzky states, LISPcraft does contain a thorough exposition of a
particular LISP dialect, namely Franz. For example, the book contains an
appendix that describes all Franz LISP functions. However, most of the book
is rather dialect independent, and major idiosyncracies are noted
throughout. The point of the thoroughness is to suggest a repetoire
of functions that programmers actually use, i. e., to convey what a real
LISP language looks like, aside from serving as a reference for Franz users
per se. As I suggest in my preface, I believe ``it is easier to learn a new
dialect having mastered another than it is having learned a language for
which there are no native speakers.''
I take strong exception to Touretzky's claim that his book offers the
``easiest route to becoming fluent in the language.'' Besides my belief in
the appropriateness of my own book for the novice, I wish to point out that
memorizing a German grammar book does NOT make one fluent in German. There
is a large body of other knowledge that is crucial to using a language
effectively, be that language natural or artificial. This fact was a prime
motivation behind my writing LISPcraft in the first place.
Rather than make the claim that my own book provides the best route to
fluency, or argue its merits as an introductory LISP text, I invite the
interested reader to judge for his or herself.
------------------------------
Date: 2 May 84 19:45:13-PDT (Wed)
From: ihnp4!oddjob!jeff @ Ucb-Vax
Subject: Re: Proposal for UUCP Project
Article-I.D.: oddjob.172
Do you suppose that when enough connections are made, the
UUCP network will spontaneously develop intelligence?
Jeff Bishop || University of Chicago
...ihnp4!oddjob!jeff || Astrology & Astrophysics Center
------------------------------
Date: 4 May 84 18:54:17-PDT (Fri)
From: hplabs!tektronix!ogcvax!sequent!richard @ Ucb-Vax
Subject: Re: Proposal for UUCP Project
Article-I.D.: sequent.483
Do you suppose that when enough connections are made, the UUCP
network will spontaneously develop intelligence?
Perhaps it already has. Maybe that's what keeps eating all those
first lines, and regurgitating the weeks-old news.
____________________________________________
The preceding should not be construed as the statement or opinion of the
employers or associates of the author. It might not even be the author's.
I try to make a point of protecting the innocent,
but none of them can be found...
...!sequent!richard
------------------------------
Date: 11 May 84 19:27:29-PDT (Fri)
From: decvax!minow @ Ucb-Vax
Subject: Re: Proposal for UUCP Project
Article-I.D.: decvax.482
An earlier discussion of this topic may be found in the story
"Inflexible Logic" by Russell Maloney (The New Yorker, 1940)
reprinted in The World of Mathematics, Vol. 4, pp. 2262-2267.
Martin Minow
decvax!minow
------------------------------
Date: 7 May 84 11:02:00-PDT (Mon)
From: pur-ee!uiucdcs!uicsl!dinitz @ Ucb-Vax
Subject: Re: Re: metaphors - (nf)
Article-I.D.: uicsl.15500034
FLAME ON
Your complaint that a comparison using "like" is a simile (and not a
metaphor) is technically correct. But it shows that you're not
following the research. Metaphor or simile (or juxtaposition, etc.),
these figures of speech raise the same problems and questions of how
analogical reasoning works, how comparisons convey meaning, how do
people dream them up, and how do other people understand them. For
this reason the word metaphor is used to refer collectively to the
whole lot of them. Pretending you're a high school English teacher
doesn't help.
FLAME OFF
------------------------------
Date: 10 May 84 21:16:05-PDT (Thu)
From: decvax!genrad!wjh12!foxvax1!brunix!jah @ Ucb-Vax
Subject: Re: Non-competition clauses
Article-I.D.: brunix.7927
You should be aware that it is not necessarily the case that you MUST
sign the non-disclosure agreement exactly as worded. I recently signed
on as consultant with a company which had a very stringent (and absolutely
ridiculous) nondisclosure/non-competition clause form. I refused to sign
certain sections (mainly those limiting me from practicing AI, consulting
for others where there was no conflict of interest, etc.) We eventually
eliminated those clauses, rewrote the contract and I signed willingly.
Similarly, another company I worked for was unwilling to change the document,
but, when I refused to sign away my rights, they pointed out that I got
to fill in a section with information about what things I already had going
for me (that is, what things I had done previously so the company had no claim
on these things). Since the company's contract included such things as
"no competing business" and the like, I was able to claim prior rights to
"artificial intelligence research", "natural language processing", and
"expert systems research." The very vagueness of these things, according
to my legal advisor, makes it that much harder for the company to really do
anything.
A final note, most companies will clain they do do this "as red tape" and
will "not really hassle you." Don't believe them! They've got more bucks
then you and if it goes to court, EVEN IF YOU WIN, it will cost you more
than you can afford. Speak to a lawyer, change contracts, etc. In the AI
world we've got a seller's market. Take advantage of it, these companies
want you, and will be willing to negotiate.
Sorry if I do go on...
Jim Hendler
------------------------------
Date: Wed 9 May 84 18:08:03-PDT
From: Dikran Karagueuzian <DIKRAN@SU-CSLI.ARPA>
Subject: Seminar - Content-Addressable Memory
[Forwarded from the CSLI Newsletter by Laws@SRI-AI.]
FOR THE RECORD
CSLI post-doctoral fellow Pentti Kanerva was a guest lecturer at MIT
Tuesday, May 1. The topic of his lecture was "Random-access Memory with a
Very Large Address Space (1 2000) as a Model of Human Memory: Theory and
Implementation." Douglas R. Hofstadter was host. Following is an abstract
of the lecture.
Humans can retrieve information from memory according to content (recalling
and recognizing previously encountered objects) and according to temporal
sequence (performing a learned sequence of actions). Retrieval times
indicate the direct retrieval of stored information.
In the present theory, memory items are represented by n-bit binary words
(points of space {0,1}n. The unifying principle of the theory is that the
address space and the datum space of the memory are the same. As in the
conventional random-access memory of a computer, any stored item can be
accessed directly by addressing the location in which the item is stored;
the sequential retrieval is accomplished by storing the memory record as
a linked list. Unlike in the conventional random-access memory, many
locations are accessed at once, and this accounts for recognition.
Three main results have been obtained: (1) The properties of neurons allow
their use as address decoders for a generalized random-access memory;
(2) distributing the storage of an item in a set of locations makes very
large address spaces (2 1000) practical; and (3) structures similar to those
suggested by the theory are found in the cerrebellum.
------------------------------
Date: 11 May 1984 07:08:26-EDT
From: Mark.Fox@CMU-RI-ISL1
Subject: IEEE AI Conf. Call for Papers
[Forwarded from the SRI bboard by Laws@SRI-AI.]
CALL FOR PAPERS
IEEE Workshop on Principles of Knowledge-Based Systems
Sheraton Denver Tex, Denver, Colorado, 3 - 4 December 1984
Purpose:
The purpose of this conference is to focus attention on the principle theories
and methods of artificial intelligence which have played an important role in
the construction of expert and knowledge-based systems. The workshop will
provide a forum for researchers in expert and knowledge-based systems to
discuss the concepts which underly their systems. Topics include:
- Knowledge Acquisition.
* manual elicitation.
* machine learning.
- Knowledge Representation.
- Causal modeling.
- The Role of Planning in Expert Reasoning
- Knowledge Utilization.
* rule-based reasoning
* theories of evidence
* focus of attention.
- Explanation.
- Validation.
* measures.
* user acceptance.
Please send eight copies of a 1000-2000 word double-space, typed, summary of
the proposed paper to:
Mark S. Fox
Robotics Institute
Carnegie-Mellon University
Pittsburgh, Pennsylvania 15213
All submissions will be read by the program committee:
- Richard Duda, Syntelligence
- Mark Fox, Carnegie-Mellon University
- John McDermott, Carnegie-Mellon University
- Tom Mitchell, Rutgers University
- John Roach, Virginia Polytechnical Institute
- Reid Smith, Schlumberger Corp.
- Mark Stefik, Xerox Parc
- Donald Waterman, Rand Corp.
Summaries are to focus primarily on new principles, but each principle should
be illustrated by its use in an knowledge-based system. It is important to
include specific findings or results, and specific comparisons with relevant
previous work. The committee will consider the appropriateness, clarity,
originality, significance and overall quality of each summary.
June 7, 1984 is the deadline for the submission of summaries. Authors will be
notified of acceptance or rejection by July 23, 1984. The accepted papers must
be typed on special forms and received by the program chairman at the above
address by September 3, 1984. Authors of accepted papers will be expected to
sign a copyright release form.
Proceedings will be distributed at the workshop and will be subsequently
available for purchase from IEEE. Selected full papers will be considered
(along with papers from the IEEE Conference on AI and Applications) for a
special issue of IEEE PAMI on knowledge-based systems to be published in Sept.
1985. The deadline for submission of full papers is 16 December 1984.
General Chairman
John Roach
Dept. of Computer Science
Virginia Polytechnic Institute
Blacksburg, VA
Program Co-Chairmen
Mark S. Fox Tom Mitchell
Robotics Institute Dept. of Computer Science
Carnegie-Mellon Univ. Rutgers University
Pittsburgh, PA New Brunswick, NJ
Registration Chairman Local Arrangements Chairman
Daniel Chester David Morgenthaler
Dept. of Computer Science Martin Marietta Corp.
University of Delaware Denver, Colorado
Newark, Delaware
------------------------------
End of AIList Digest
********************
