Copy Link
Add to Bookmark
Report
AIList Digest Volume 1 Issue 051
AIList Digest Wednesday, 31 Aug 1983 Volume 1 : Issue 51
Today's Topics:
Expert Systems - Availability & Dissent,
Automatic Translation - State of the Art,
Fifth Generation - Book Review & Reply
----------------------------------------------------------------------
Date: 26 Aug 83 17:00:18-PDT (Fri)
From: decvax!ittvax!dcdwest!benson @ Ucb-Vax
Subject: Expert Systems
Article-I.D.: dcdwest.216
I would like to know whether there are commercial expert
systems available for sale. In particular, I would like to
know about systems like the Programmer's Apprentice, or other
such programming aids.
Thanks in advance,
Peter Benson
!decvax!ittvax!dcdwest!benson
------------------------------
Date: 26 Aug 83 11:12:31-PDT (Fri)
From: decvax!genrad!mit-eddie!rh @ Ucb-Vax
Subject: bulstars
Article-I.D.: mit-eddi.656
from AP (or NYT?)
COMPUTER TROUBLESHOOTER:
'Artificially Intelligent' Machine Analyses Phone Trouble
WASHINGTON - Researchers at Bell Laboratories say
they've developed an ''artificially intelligent'' computer
system that works like a highly trained human analyst to
find troublespots within a local telephone network. Slug
PM-Bell Computer. New, will stand. 670 words.
Oh, looks like we beat the Japanese :-( Why weren't we told that
'artificial intelligence' was about to exist? Does anyone know if
this is the newspaper's fault, or if the guy they talked to just
wanted more attention???
-- Randwulf
(Randy Haskins);
Path= genrad!mit-eddie!rh
or... rh@mit-ee (via mit-mc)
------------------------------
Date: Mon 29 Aug 83 21:36:04-CDT
From: Jonathan Slocum <LRC.Slocum@UTEXAS-20.ARPA>
Subject: claims about "solving NLP"
I have never been impressed with claims about "solving the Natural
Language Processing problem" based on `solutions' for 1-2 paragraphs
of [usu. carefully (re)written] text. There are far too many scale-up
problems for such claims to be taken seriously. How many NLP systems
are there that have been applied to even 10 pages of NATURAL text,
with the full intent of "understanding" (or at least "treating in the
identical fashion") ALL of it? Very few. Or 100 pages? Practically
none. Schank & Co.'s "AP wire reader," for example, was NOT intended
to "understand" all the text it saw [and it didn't!], but only to
detect and summarize the very small proportion that fell within its
domain -- a MUCH easier task, esp. considering its miniscule domain
and microscopic dictionary. Even then, its performance was -- at best
-- debatable.
And to anticipate questions about the texts our MT system has been
applied to: about 1,000 pages to date -- NONE of which was ever
(re)written, or pre-edited, to affect our results. Each experiment
alluded to in my previous msg about MT was composed of about 50 pages
of natural, pre-existing text [i.e., originally intended and written
for HUMAN consumption], none of which was ever seen by the project
linguists/programmers before the translation test was run. (Our
dictionaries, by the way, currently comprise about 10,000 German
words/phrases, and a similar number of English words/phrases.)
We, too, MIGHT be subject to further scale-up problems -- but we're a
damned sight farther down the road than just about any other NLP
project has been, and have good reason to believe that we've licked
all the scale-up problems we'll ever have to worry about. Even so, we
would NEVER be so presumptuous as to claim to have "solved the NLP
problem," needing only a large collection of `linguistic rules' to
wrap things up!!! We certainly have NOT done so.
REALLY, now...
------------------------------
Date: Mon 29 Aug 83 17:11:26-CDT
From: Jonathan Slocum <LRC.Slocum@UTEXAS-20.ARPA>
Subject: Machine Translation - a very short tutorial
Before proclaiming the impossibility of automatic [i.e., computer]
translation of human languages, it's perhaps instructive to know
something about how human translation IS done -- and is not done -- at
least in places where it's taken seriously. It is also useful,
knowing this, to propose a few definitions of what may be counted as
"translation" and -- more to the point -- "useful translation."
Abbreviations: MT = Machine Translation; HT = Human Translation.
To start with, the claim that "a real translator reads and understands
a text, and then generates [the text] in the [target] language" is
empty. First, NO ONE really has anything like a good idea of HOW
humans translate, even though there are schools that "teach
translation." Second, all available evidence indicates that (point #1
notwithstanding), different humans do it differently. Third, it can
be shown (viz simultaneous interpreters) that nothing as complicated
as "understanding" need take place in all situations. Fourth,
although the contention that "there generally aren't 1-1
correspondences between words, phrases..." sounds reasonable, it is
in fact false an amazing proportion of the time, for languages with
similar derivational histories (e.g., German & English, to say nothing
of the Romance languages). Fifth, it can be shown that highly
skilled, well-respected technical-manual translators do not always (if
ever) understand the equipment for which they're translating manuals
[and cannot, therefore, be argued to understand the original texts in
any fundamentally deep sense] -- and must be "understanding" in a
shallower, probably more "linguistic" sense (one perhaps more
susceptible to current state-of-the-art computational treatment).
Now as to how translation is performed in practice. One thing to
realize here is that, at least outside the U.S. [i.e., where
translation is taken seriously and where almost all of it is done], NO
HUMAN performs "unrestricted translation" -- i.e., human translators
are trained in (and ONLY considered competent in) a FEW AREAS.
Particularly in technical translation, humans are trained in a limited
number of related fields, and are considered QUITE INCOMPETENT outside
those fields. Another thing to realize is that essentially ALL
TRANSLATIONS ARE POST-EDITED. I refer here not to stylistic editing,
but to editing by a second translator of superior skill and
experience, who NECESSARILY refers to the original document when
revising his subordinate's translation. The claim that MT is
unacceptable IF/BECAUSE the results must be post-edited falls to the
objection that HT would be unacceptable by the identical argument.
Obviously, HT is not considered unacceptable for this reason -- and
therefore, neither should MT. All arguments for acceptablility then
devolve upon the question of HOW MUCH revision is necessary, and HOW
LONG it takes.
Happily, this is where we can leave the territory of pontifical
pronouncements (typically utterred by the un- or ill-informed), and
begin to move into the territory of facts and replicable experiments.
Not entirely, of course, since THERE IS NO SUCH THINGS AS A PERFECT
TRANSLATION and, worse, NO ONE CAN DEFINE WHAT CONSTITUTES A GOOD
TRANSLATION. Nevertheless, professional post-editors are regularly
saddled with the burden of making operational decisions about these
matters ("Is this sufficiently good that the customer is likely to
understand the text? Is it worth my [company's] time to improve it
further?"). Thus we can use their decisions (reflected, e.g., in
post-editing time requirements) to determine the feasibility of MT in
a more scientific manner; to wit: what are the post-editing
requirements of MT vs. HT? And in order to assess the economic
viability of MT, one must add: taking all expenses into account, is MT
cost-effective [i.e., is HT + human revision more or less expensive
than MT + human revision]?
Re: these last points, our experimental data to date indicate that (1)
the absolute post-editing requirements (i.e., something like "number
of changes required per sentence") for MT are increased w.r.t. HT
[this is no surprise to anyone]; (2) paradoxically, post-editing time
requirements of MT is REDUCED w.r.t. HT [surprise!]; and (3) the
overall costs of MT (including revision) are LESS than those for HT
(including revision) -- a significant finding.
We have run two major experiments to date [with our funding agency
collecting the data, not the project staff], BOTH of which produced
these results; the more recent one naturally produced better results
than the earlier one, and we foresee further improvements in the near
future. Our finding (2) above, which SEEMS inconsistent with finding
(1), is explainable with reference to the sociology of post-editing
when the original translator is known to be human, and when he will
see the results (which probably should, and almost always does,
happen). Further details will appear in the literature.
So why haven't you heard about this, if it's such good news? Well,
you just did! More to the point, we have been concentrating on
producing this system more than on writing papers about it [though I
have been presenting papers at COLING and ACL conferences], and
publishing delays are part of the problem [one reason for having
conferences]. But more papers are in the works, and the secret will
be out soon enough.
------------------------------
Date: 26 Aug 83 1209 PDT
From: Jim Davidson <JED@SU-AI>
Subject: Fifth Generation (Book Review)
[Reprinted from the SCORE BBoard.]
14 Aug 8
by Steven Schlossstein
(c) 1983 Dallas Morning News (Independent Press Service)
THE FIFTH GENERATION: Artificial Intelligence and Japan's Computer
Challenge to the World. By Edward Feigenbaum and Pamela McCorduck
(Addison-Wesley, $15.55).
(Steven Schlossstein lived and worked in Japan with a major Wall
Street firm for more than six years. He now runs his own Far East
consulting firm in Princeton, N.J. His first novel, ''Kensei,-' which
deals with the Japanese drive for industrial supremacy in the high
tech sector, will be published by Congdon & Weed in October).
''Fukoku Kyohei'' was the rallying cry of Meiji Japan when that
isolated island country broke out of its self-imposed cultural cocoon
in 1868 to embark upon a comprehensive plan of modernization to catch
up with the rest of the world.
''Rich Country, Strong Army'' is literally what is meant.
Figuratively, however, it represented Japan's first experimentation
with a concept called industrial policy: concentrating on the
development of strategic industries - strategic whether because of
their connection with military defense or because of their importance
in export industries intended to compete against foreign products.
Japan had to apprentice herself to the West for a while to bring
it off.
The military results, of course, were impressive. Japan defeated
China in 1895, blew Russia out of the water in 1905, annexed Korea and
Taiwan in 1911, took over Manchuria in 1931, and sat at the top of the
Greater East Asia Co-Prosperity Sphere by 1940. This from a country
previously regarded as barbarian by the rest of the world.
The economic results were no less impressive. Japan quickly became
the world's largest shipbuilder, replaced England as the world's
leading textile manufacturer, and knocked off Germany as the premier
producer of heavy industrial machinery and equipment. This from a
country previously regarded as barbarian by the rest of the world.
After World War II, the Ministry of Munitions was defrocked and
renamed the Ministry of International Trade and Industry (MITI), but
the process of strategy formulation remained the same.
Only the postwar rendition was value-added, and you know what
happened. Japan is now the world's No. 1 automaker, produces more
steel than anyone else, manufactures over half the TV sets in the
world, is the only meaningful producer of VTRs, dominates the 64K
computer chip market, and leads the way in one branch of computer
technology known as artificial intelligence (AI). All this from a
country previously regarded as barbarbian by the rest of the world.
What next for Japan? Ed Feigenbaum, who teaches computer science
at Stanford and pioneered the development of AI in this country, and
Pamela McCorduck, a New York-based science writer, write that Japan is
trying to dominate AI research and development.
AI, the fifth generation of computer technology, is to your
personal computer as your personal computer is to pencil and paper. It
is based on processing logic, rather than arithmetic, deals in
inferences, understands language and recognizes pictures. Or will. It
is still in its infancy. But not for long; last year, MITI established
the Institute for New Generation Computer Technology, funded it
aggressively, and put some of the country's best brains to work on AI.
AI systems consist of three subsystems: a knowledge base needed
for problem solving and understanding, an inference subsystem that
determines what knowledge is relevant for solving the problem at hand,
and an interaction subsystem that facilitates communication between
the overall system and its user - between man and machine.
Now America does not have a MITI, does not like industrial policy,
has not created an institute to work on AI, and is not even convinced
that AI is the way to go. But Feigenbaum and McCorduck argue that even
if the Japanese are not successful in developing the fifth generation,
the spin-off from this 10-year project will be enormous, with
potentially wide applications in computer technology,
telecommunications, industrial robotics, and national defense.
''The Fifth Generation'' walks you through AI, how and why Japan
puts so much emphasis on the project, and how and why the Western
nations have failed to respond to the challenge. National defense
implications alone, the authors argue, are sufficient to justify our
taking AI seriously.
Smart bombs and laser weapons are but advanced wind-up toys
compared with the AI arsenal of the future. The Pentagon has a little
project called ARPA - Advanced Research Projects Agency - that has
been supporting AI small-scale, but not with the people or funding the
authors feel is meaningful.
Unfortunately, ''The Fifth Generation'' suffers from some
organizational defects. You don't really get into AI and how its
complicated systems operate until you're almost halfway through the
book. And the chapter on industrial policy - from which all
technological blessings flow - is only three pages long. It's also at
the back of the book instead of up front, where it belongs.
But the issues are highlighted well by experts who are not only
knowledgeable about AI but who are concerned about our lack of
response to yet another challenge from Japan. The author's depiction
of the drivenness of the Japanese is especially poignant. It all boils
down to national survival.
Japan no longer is in a position of apprenticeship to the West.
[Begin garbage]
The D B LD LEAJE OW IN A EMBARRUSSINOF STRATEGIC INDUSDRIES. EAgain1u
2, with few exceptions and shampoo, but it's not trying harder - if at
all.
[End garbage]
mount an effective reaponse to the Japanese challenge? ''The
Fifth Generation'' doesn't think so, and for compelling reasons. Give
it a read.
END
------------------------------
Date: Fri 26 Aug 83 15:40:16-PDT
From: Richard Treitel <TREITEL@SUMEX-AIM>
Subject: Re: Fifth Generation (Book Review)
[Reprinted from the SCORE BBoard.]
Anybody who says the Japanese are *leading* in "one branch of computer
technology known as artificial intelligence" is out to lunch. And by
what standards is DARPA describable as small? And what is all this
BirdSong about other countries failing to "respond to the challenge"?
Hasn't this turkey read the Alvey report? Hasn't he noticed France's
vigorous encouragement of their domestic computer industry? Who in
America is not "convinced that AI is the way to go" (this was true of
the leadership in Britain until the Alvey report came out, I admit)
and what are they doing to hinder AI work? Does he think 64k RAMs are
the only things that go into computers? Does he, incidentally, know
that AI has had plenty of pioneers outside of the HPP?
More to the point, most of you know about the wildly over-optimistic
promises that were made in the 60's on behalf of AI, and what happened
in their wake. Whipping up public hysteria is a dangerous game,
especially when neither John Q. Public nor Malcolm Forbes himself can
do very much about the 5GC project, except put pressure on the local
school board to teach the kids some math and science.
- Richard
------------------------------
End of AIList Digest
********************
