Copy Link
Add to Bookmark
Report
AIList Digest Volume 2 Issue 031
AIList Digest Wednesday, 14 Mar 1984 Volume 2 : Issue 31
Today's Topics:
Humor - Who is Tom Benson?,
Linguistics - And as a Non-Logical Conjuction,
Brain Theory - Parallelism,
Seminars - Procedural Learning (Boston) &
Theorem Proving in the Stanford Pascal Verifier,
Conference - 4th Conf. on FST&TCS,
Review - Western Culture in the Computer Age
----------------------------------------------------------------------
Date: Mon 12 Mar 84 18:48:05-PST
From: ROBINSON@SRI-AI.ARPA
Subject: Tom Benson
Who is this Tom Benson and what is he doing out of
captivity?
------------------------------
Date: 12 Mar 84 11:32:17-PST (Mon)
From: hplabs!hao!seismo!rochester!stuart @ Ucb-Vax
Subject: And as a non-logical conjuction - Request for pointers
Article-I.D.: rocheste.5580
From: Stuart Friedberg <stuart>
I am looking for pointers into the linguistic and natural language
processing literature concerning the use of "and" in English as a
non-logical conjunction. That is, the use of "and" often implies
temporal sequence and/or causality. There is also a use introducing
a verb complement.
"Sheila took the ball and ran with it."
"The lights were off and I couldn't see."
"I will try and find the missing book."
I understand that treatment of conjunction and ellipsis is difficult.
Pointers to books, articles, theses, diatribes, etc. that (have sections
that) deal with "and" in this extra-logical sense will be *much* more
useful than pointers to more general treatments of conjunction and
ellipsis.
Useful things to know that I don't:
What are (all?) the senses in which "and" may be used?
Do all these interpretations apply to clause conjunction
only? (Ie, not to noun conjunction, adverb
conjunction, etc.)
What knowledge is needed/useful to determine the sense of
"and" in a given English sentence? (Given a knowledge
of all the senses of "and", how to we eliminate some
of them in a particular context?)
Is it possible to expand Ross's constraints in a reasonable
way to handle this kind of conjuction? (Constraints
on variables in syntax, thesis, MIT, 1967, etc.)
I have a few pointers already, but my only real linguistic source is
several years old. I assume that additional work has been done from
both linguistic and AI points of view. I am starting from:
1) Susan F. Schmerling, "Asymmetric Conjuction and Rules of Conversation",
in Syntax and Semantics, Vol. 3 (Speech Acts), Cole and Morgan (eds.),
Academic Press, New York, 1975
2) Stan C. Kwasny, "Treatment of Ungrammatical and Extra-grammatical
Phenomena in Natural Language Understanding Systems", Indiana University
Linguistic Club, Bloomington, IN, 1980
Stu Friedberg
{seismo, allegra}!rochester!stuart UUCP
stuart@rochester ARPA
Dept. Computer Science MAIL
University of Rochester
Rochester, NY 14627
------------------------------
Date: Tue, 13 Mar 84 18:05 EST
From: Ives@MIT-MULTICS.ARPA
Subject: Brain Theory - Parallelism
A strikingly clear picture of brain parallelism at the gross anatomical
level was presented during a lecture at MIT on the architecture of the
cerebral cortex by a neuroanatomist (Dr. Deepak Pandya, Bedford
Veterans Administration Hospital, Bedford, MA).
Almost a hundred years ago, dye studies showed that the cerebral cortex
is not a random mass of neurons, and it was mapped into a few dozen
areas, differentiated by microstructure. Later, it was shown that
lesions in a certain area always produced the same behavioral
deficiencies. Now, they have mapped out the interconnections between
the areas. The map looks like a plate of spaghetti but, when
transformed into a schematic, reveals simplicity and regularity.
Each half of the brain includes six sets of areas. Each set has a
somatic area, a visual area and an auditory area. Each area in a set
connects to the other two, forming a triangle. The six sets form a
stack because each area is connected to the area of the same kind in the
next set. The eighteen areas schematicized by this simple triangular
stack include most of the tissue in a cerebral cortex.
If I remember correctly, all mammals have this architecture. It was
surmised that one set evolved first and was replicated six times,
because the neuronal microstructure varies gradually with increasing
level. He also suggested that higher levels might process higher levels
of abstraction.
-- Jeffrey D. Ives
------------------------------
Date: 12 Mar 1984 13:39 EST (Mon)
From: "Daniel S. Weld" <WELD%MIT-OZ@MIT-MC.ARPA>
Subject: Procedural Learning (Boston)
[Forward from the MIT bboard by SASW@MIT-MC.]
Wednesday, March 14 4:00pm 8th floor playroom
Acquisition of Procedural Knowledge from Examples
P. M. Andreae
I will describe NODDY - a system that acquires procedures from
examples. NODDY is a variation of concept learning in which the
concepts to be learned are procedures in the form of simple robot
programs. The procedures are acquired by generalising examples
obtained by leading a robot through a sequence of steps. Three
distinct types of generalisation are involved: structure
generalisation (eg. loops and branches), event generalisation (eg. the
branching conditions), and function induction.
I will also discuss two principles that arise out
of, and are illustrated by, NODDY. I claim that these principles have
application, not only to procedure acquisition, but also to any system
that does partial matching and/or generalisation of any kind.
------------------------------
Date: Mon 12 Mar 84 19:16:47-PST
From: Richard Treitel <TREITEL@SUMEX-AIM.ARPA>
Subject: Theorem Proving in the Stanford Pascal Verifier
[Forwarded from the Stanford bboard by Laws@SRI-AI.]
The Automatic Inference Seminar will meet on Wednesday March 14th in MJH 352
(note change of room from 301) at 1:30 p.m.
(This is tax-filing season; I'm getting slightly too many groanworthy remarks
about "automatic deduction", hence the name change).
Speaker: Richard Treitel (oh no, not again)
Subject: Theorem Proving in the Stanford Pascal Verifier
The Stanford Pascal Verifier was developed in the late 1970's for research in
program verification. Its deductive component, designed mainly by Greg Nelson
and Derek Oppen, has some features not found in many other natural deduction
systems, including a powerful method for dealing with equalities, a general
framework for combining the results of decision procedures for fragments of the
problem domain, and a control structure based on an unusual "normal form" for
expressions. I will attempt to explain these and relate them both to other
provers and to post-Oppen work on the same technology.
------------------------------
Date: 9 Mar 84 9:52:21-PST (Fri)
From: harpo!ulysses!burl!clyde!akgua!psuvax!narayana @ Ucb-Vax
Subject: Call for 4th Conf. FST&TCS Bangalore India Dec 13-15.
Article-I.D.: psuvax.815
Subject: Call for papers
4th Conference on Foundations of Software
Engineering and Theoretical Computer Science
Bangalore, INDIA, DECEMBER 13-15, 1984
Sponsor: Tata Institute of Fundamental Research, Bombay, India.
Conference advisory committe:
A.Chandra(IBM), B.Chandrasekharan(Ohio state), S.Crespi Reghizzi(Milan)
D.Gries(Cornell), A.Joshi(Penn), U.Montanari(Pisa), J.H.Morris(CMU),
A.Nakamura(Hiroshima), R.Narasimhan(TIFR), J.Nivergelt(ETH), M.Nivat(Paris)
R.Parikh(Ney York), S.R.Kosaraju(Johns Hopkins), B.Reusch(Dortmund),
R.Sethi(Bell labs), S.Sahni(Minnesota), P.S.Tiagarajan(Aarhus),
W.A.Wulf(Tartan labs).
Papers are invited in the following areas:
Programming languages and systems
Program correctness and proof methodologies
Formal semantics and specifications
Theory of computation
Formal languages and automata
Algorithms and complexity
Data bases
Distributed computing
Computing practice
Papers will be REFEREED and a final selection will be made by the programme
committe.
Authors should send four copies of each paper to
Chairman, FST&TCS Programme Committe
Tata Institute of Fundamental Research
Homi Bhabha Road, BOMBAY, 400 005, India
Due date for receiving full papers: MAY 31, 1984.
Authors will be notified of acceptance/rejection by: JULY 31,1984
Camera ready papers must be submitted by: SEP 15,1984
PROCEEDINGS WILL BE PUBLISHED. For further details contact the above address.
Programme Committe: M.Joseph(TIFR), S.N.Maheswari(IIT), S.L.Mahindiratta(IIT),
K.V.Nori(Tata RDDC), S.V.Rangaswamy(IISC), R.K.Shyamasundar
(TIFR), R.Siromani(Madras Christian college).
------------------------------
Date: 12 Mar 84 2053 PST
From: Frank Yellin <FY@SU-AI.ARPA>
Subject: Western Culture in the Computer Age
[Forwarded from the Stanford bboard by Laws@SRI-AI.]
n066 1529 12 Mar 84
BC-BOOK-REVIEW Undated
(Culture)
By CHRISTOPHER LEHMANN-HAUPT
c.1984 N.Y. Times News Service
TURING'S MAN. Western Culture in the Computer Age. By J. David
Bolter. 264 pages. University of North Carolina Press. Hard-cover,
$19.95; paper, $8.95.
J. David Bolter, the author of ''Turing's Man: Western Culture in
the Computer Age,'' is both a classicist who teaches at the
University of North Carolina and a former visiting fellow in computer
science at Yale University. This unusual combination of talents may
not qualify him absolutely to offer a humane view of the computer
age, or what he refers to as the age of Turing's man, after Alan M.
Turing, the English mathematician and logician who offered early
theoretical descriptions of both the computer and advanced artificial
intelligence.
But his two fields of knowledge certainly provide Bolter with an
unusual perspective on contemporary developments that many observers
fear are about to usher in an age of heartless quantification, if not
the final stages of Orwellian totalitarianism. In Bolter's view,
every important era of Western civilization has had what he calls its
''defining technology'' which ''develops links, metaphorical or
otherwise, with a culture's science, philosophy, or literature; it is
always available to serve as a metaphor, example, model, or symbol.''
To the ancient Greeks, according to Bolter, the dominant
technological metaphor was the drop spindle, a device for twisting
yarn into thread. Such a metaphor implied technology as a controlled
application of power. To Western Europe after the Middle Ages, the
analogues to the spindle were first, the weight-driven clock, a
triumph of mechanical technology, and later, the steam engine, a
climax of the dynamic. In Bolter's subtly developed argument, the
computer - obviously enough the present age's defining metaphor - is
an outgrowth of both the clock and the steam engine. Yet,
paradoxically, the computer also represents a throwback.
Everything follows from this. In a series of closely reasoned
chapters on the way in which the computer has redefined our notions
of space, time, memory, logic, language and creativity, Bolter
reviews a subtle but recuing pattern in which the computer
simultaneously climaxes Western technology and returns us to ancient
Greece. He concludes that if the ancient ideals were balance,
proportion and handicraft (the use of the spindle), and the Western
European one was the Faustian quest for power through knowledge
(understanding a clockwork universe to attain the dynamism of the
steam engine), then Turing's man combines the two.
''In his own way, computer man retains and even extends the Faustian
tendency to analyze,'' Bolter concludes. ''Yet the goal of Faustian
analysis was to understand, to 'get to the bottom' of a problem,''
whereas ''Turing's man analyzes not primarily to understand but to
act.''
He continues: ''For Turing's man, knowledge is a process, a skill,''
like the ancient arts of spinning or throwing a pot. ''A man or a
computer knows something only if he or it can produce the right
answer when asked the right question.'' Faustian depth ''adds nothing
to the program's operational success.''
Now in portraying Turing's man, Bolter may seem to be overburdening
a few simple metaphors. Yet his argument is developed with remarkable
concreteness. Indeed, if his book has any fault, it lies in the
extent to which he has detailed the slightly repetitious and
eventually predictable pattern of argument described above.
Yet what is far more important about ''Turing's Man'' is its success
in bridging the gap between the sciences and the humanities. I can
only guess at how much it will inform the computer technologist about
philosophy and art, but I can vouch for how much it has to say to the
nonspecialist about how computers work. The inaccessibility of the
computer's inner functioning may well be a key to the author's case
that Turing's man is returning to the ancient Greek's satisfaction in
the surface of things, but after reading Bolter's book, this reader
found the computer far less mysterious. Not incidentally, the book
makes us understand why computers aren't really all that good at
doing mathematics (they can't get a grip on the notion of infinity);
and it far surpasses Andrew Hodges's recent biography of Alan Turing
in explaining Turing's Game for testing artificial intelligence.
But most provocative about this study is what it has to say about
the political implications of the computer age. Will Turing's man
prove the instrument of Orwell's Big Brother, as so many observers
are inclined to fear? Very likely not, says Bolter:
''Lacking the intensity of the mechanical-dynamic age, the computer
age may in fact not produce individuals capable of great good or
evil. Turing's man is not a possessed soul, as Faustian man so often
was. He does not hold himself and his world in such deadly earnest;
he does not speak of 'destiny' but rather of 'options.' And if the
computer age does not produce a Michelangelo and a Goethe, it is
perhaps less likely to produce a Hitler or even a Napoleon. The
totalitarian figures were men who could focus the Faustian commitment
of will for their ends. What if the will is lacking? The premise of
Orwell's '1984' was the marriage of totalitarian purpose with modern
technology. But the most modern technology, computer technology, may
well be incompatible with the totalitarian monster, at least in its
classic form.''
Indeed, according to Bolter, Turing's man may be more inclined to
anarchy than to totalitarianism. This may be whistling past the
graveyard. But in Bolter's stimulating analysis, it also makes a kind
of homely sense.
------------------------------
End of AIList Digest
********************
