LAWS@SRI-AI.ARPA (05/19/85)
From: AIList Moderator Kenneth Laws <AIList-REQUEST@SRI-AI>
AIList Digest Sunday, 19 May 1985 Volume 3 : Issue 66
Today's Topics:
Query - Exploring Language with Logo,
Binding - Mark Grover,
Tools - STATECHARTS Reference & Lisp Machines & TI's Explorer,
Opinion - Introspection and Communication & Emotional Reasoning,
Journal - AI in Engineering
----------------------------------------------------------------------
Date: 18 May 85 14:48 EDT
From: Bill Caswell <caswell@nswc-wo>
Subject: Exploring Language with Logo
Last November, a seminar on material from a forthcoming book, "Exploring
Language with Logo", by Paul Goldenberg and Wallace Feurzeig, was announced
in this list. The book was to be published 'first quarter, 1985' by Harper
and Row, and I have not seen anything about it since. Has it been published,
or is it otherwise available? Bill
------------------------------
Date: 15 May 1985 1951-EDT (Wednesday)
From: trwatf!maverick@seismo.ARPA (Mark D. Grover)
Subject: Binding - Mark Grover
Binding: Mark D. Grover
Formerly: TRW Defense Systems Group (Advanced Technology Facility)
Fairfax, Virginia [trwatf!maverick@SEISMO]
Beginning 5/24/85: Advanced Information & Decision Systems
Washington technical office [Arlington, Virginia]
New ARPAnet address: GROVER@AIDS-UNIX
------------------------------
Date: Thu, 16 May 85 18:43:10 edt
From: Walter Hamscher <hamscher@MIT-HTVAX.ARPA>
Subject: STATECHARTS reference
Here's the reference:
Harel, D.
Statecharts: A Visual Approach to Complex Systems
CS84-05, Department of Applied Math, Weizmann Institute of Science,
Rehovot, Israel. 1984. 35pp.
Harel's developed a nice notation for complex finite-state machines
by making use of hierarchic states, allowing both XOR-decomposition
of a state (i.e. the system's in one & only one of the substates)
and AND-decomposition (i.e. the system's in a state defined by the
cross product of the substates). By defining some other conventions
about transition arcs, the result is surprisingly compact and
expressive. However, his discussion of adding actions to arcs and
states seemed a bit less elegant. Worth a look.
Incidentally, Harel's netadress is harel%wisdom@WISCVM.ARPA ...
------------------------------
Date: 17 May 85 08:27 PDT
From: Masinter.pa@Xerox.ARPA
Subject: Lisp Machines
In reply to curtis L. Goodhart <goodhart%cod@Nosc>
You don't need any specific kind of machine to run lisp. There are lisp
implementations for lots of machines. Some of them can be quite fast.
Specialized instruction sets for Lisp allow you to get performance
without declarations, to retain full information for symbolic debugging,
better garbage collection performance, and the ability to retain
run-time type-checking even in performance critical loops.
Memory utilization is generally much better; for example, the last time
I measured, Interlisp-D programs took less than half the bytes to
represent than the same programs on a VAX, and CDR-coding allowed the
average size for a CONS cell to be 36 bits rather than 64. Lisp machines
can afford to use compact data structures because the instruction set
can be designed to deal with them; "conventional" instruction sets
generally cannot be skewed to make otherwise obscure manipulations
(fetch bits 4-5 and branch if 0) into single operations.
All of the lisp machines on the market come with software as well as
hardware. Dedicated machines allow the implementation of integrated
environments and high performance graphical interfaces which are
generally unavailable on conventional computers. The software
enviornment can be major portion of the "value added" of the system; the
proportion is likely to continue as hardware costs drop.
What is true of larger machines today will be true of smaller ones in
the future; while prices of all computer components drop, so do the
prices that people expect to pay for them.
The earliest reference I've found is Peter Deutsch's paper, "A Lisp
machine with very compact programs," IJCAI 1973. "Programming
environments" are quite popular these days, with whole conferences
devoted to them, although most of the papers spend their time extolling
individual environments rather than the notion of them.
There are other justifications more recent found in the marketing
literature of various vendors, written to make sure that its clear that
each is the best.
------------------------------
Date: Sat 18 May 85 12:17:18-CDT
From: Werner Uhrig <CMP.WERNER@UTEXAS-20.ARPA>
Subject: TI's Explorer won Industrial Design Awards
[ from the Austin American Statesman - May 18 ]
TEXAS INSTRUMENTS' AUSTIN-developed EXPLORER artificial intelligence
workstation has received one of 23 Industrial Design Awards given by
ID Magazine. More than 800 entries were judged on innovation,
problem-solving, aesthetics, materials and use of the products.
The computer also won a design award in a competition at the
Hannover Trade Fair in West Germany.
------------------------------
Date: Tue, 14 May 85 9:28:33 EDT
From: Bruce Nevin <bnevin@BBNCCH.ARPA>
Subject: indignation and clarity
From: _Bob <Carter@RUTGERS.ARPA> (the Midnight Theorizer)
I have some doubts about that part of AI which asserts the validity
of scientific inquiry that gathers "data" by introspection. But I
have even greater doubts about moral indignation as a criterion for
rejection of hypotheses (or of humor, for that matter).
Chomsky has argued that introspective data are necessary for a science
of language. His followers and apostates have demonstrated how
hazardous they can be. His teacher, Zellig Harris, has shown that
appeals to introspective data can be well defined and few with no loss
to linguistic science.
As to moral indignation, it is possible to read this message and wonder
whether it is allowed any legitimacy, and to ask under what circumstances
the writer would find moral indignation appropriate. I believe this would be a
misreading of the message, but nonetheless a possible reading.
Ambiguity is ineluctable. I believe we have to anticipate, insofar as
is possible, the ways in which we may be misconstrued, and make our
communications clear IN TERMS APPROPRIATE TO THE MISCONSTRUING AUDIENCE.
To insist that the burden of rightly understanding rests alone on those
who have misunderstood is practically to guarantee a failure of
communication.
There is another well-known mechanism in which a social level of
communication is incongruent with a psychological level. A simple
example is the stereotype that has been made of Bateson's work on the
double bind (Mother says `I love you, come here' while appearing fearful
and angry, or in a context that punishes the child). The communication
that is not consciously acceptable to the sender is kept out of the
sender's awareness. The receiver must choose which message to believe
and act on consciously, and believes and acts on the other message
unconsciously.
Most miscommunication, it has been argued, occurs when people make
different choices about which parts of the complex message are socially
acceptable and which are not. And people often vigorously resist
recognizing the beliefs and disbeliefs that they have been communicating
and otherwise acting on out of conscious awareness.
The `talking past each other' that we have just seen on this list
concerning rape and humor has this character.
By the way, I recommend Milton Rokeach, The Open and Closed Mind,
for a perspective on belief-disbelief systems that would be useful for
modelling human `knowledge bases'.
Bruce Nevin (bn@bbncch.arpa)
------------------------------
Date: Fri, 17 May 85 04:52:50 est
From: "Marek W. Lugowski" <marek%indiana.csnet@csnet-relay.arpa>
Subject: Emotion & AI and its discussion on the AIList
The discussion of emotion in the context of AI can be of benefit to us all.
For too long, AI has equated thinking with "reasoning", where reasoning is
a something, and entirely apart from a something else called "passion". This
dichotomy is Plato's, and likely an AI liability, given the affective basis
for one's actions and one's interpretation of actions. Judgement as such
crucially depends on emotional state, as psychiatrists well know. Judgment
is what yesterday's Shakey the Robot lacked, and what today's naive physics
is ostentatiously after. Alas, nothing's changed: Each confines itself to
the rigid minuet of "reasoning" devoid of emotional context.
Incidentally, the AIList-Digest's discussion of emotion has been stirring up
ideas in the classical vein of emotional muckraking, not so much dispassionate
analysis. What's needed is computational ideas on how to model emotions: to
start with, in a "toy" domain. And don't assume that they are if-then rules!
-- Marek Lugowski
marek@indiana.csnet
Indiana U. CS Department
Bloomington 47405
------------------------------
Date: Monday, 13 May 1985 23:34:24 EDT
From: Duvvuru.Sriram@cmu-ri-cive.arpa
Subject: New Journal - AI in Engineering
A new journal on the applications of AI in Engineering will be launched at
the First Conference on Applications of AI in Engineering next April.
Details of the journal are provided below. The deadline for sending papers
for the first issue is November 15, 1985.
International Journal for
ARTIFICIAL INTELLIGENCE IN ENGINEERING
Editors:
D. Sriram Dr. K. J. MacCallum
Dept. Civil Engineering Dept. of Ship & Marine Tech.
Carnegie-Mellon University Marine Technology Centre
Pittsburgh, PA 15213 100 Montrose Street, Glasgow
USA Scotland
Recent advances in the applications of artificial intelligence (AI)
are begining to make a major impact in the world of engineering. As a
consequence, important opportunities for using computers to tackle
engineering problems in new ways are opening up. The engineering
industry must keep itself fully informed of these developments and
contribute to them, if it is to realize the potential of these
advances.
The aim of this journal is to establish a forum for a fruitful
exchange of ideas through the publication of up-to-date research
results and recent developments. While the range of AI topics covered
by the journal is broadly based, the emphasis will be on research and
development leading to problem solving. The journal should appeal to
engineers for all disciplines who are involved in research,
development or implementation of computer systems.
Members of the journal's leading international Editorial Board will be
responsible for the review of all papers submitted for publication to
ensure that readers receive a consistently high quality of work in
each issue.
>From time to time review papers will be published to provide
state-of-the-art analyses of various areas of current interest. In
addition the journal will provide a review of books and reports in
this expanding field as well as news letters, and a dairy of events of
conferences, courses and meetings in AI.
FIELDS COVERED
- Expert systems
- Knowledge representation
- Knowledge-based simulation
- Computer aided design
- Design modelling
- Cognitive modelling of engineering problems
- Learning
- Computer based training
- Intelligent tutors
- Robotics
- Planning and scheduling
- Constraint management
- Natural language applications
- Database interfaces
- Graphical interfaces
- Computer integrated manufacturing
REGULAR FEATURES
- Discussion of published papers
- Conference and meeting reports
- Personalia
- Letters
- Book reviews
- Current literature
- Calendar of events
- News
POLICY AND CALL FOR PAPERS
The editorial policy encourages the publication of research articles
on recent advances in AI in engineering. State-of-the-art papers will
also be included from time to time. Details on presentation of papers
for consideration of publication can be obtained from the editors.
SUBSCRIPTION
This journal will keep you abreast with the major research and
development work around the world in AI applied to engineering
problems. The editors and their international Editorial Board are well
placed to keep you informed in these rapidly developing field. For
more details contact:
Lance Sucharov
Publishing Director
CML Publications
Ashurst Lodge
Ashurst
Hants SO4 2AA
England
For a author's tookit (in US), write to D. Sriram or send mail to
sriram@cmu-ri-cive.arpa.
[The subscription price has not yet been determined. -- KIL]
------------------------------
End of AIList Digest
********************