Copy Link
Add to Bookmark
Report
AIList Digest Volume 4 Issue 165
AIList Digest Thursday, 10 Jul 1986 Volume 4 : Issue 165
Today's Topics:
Books - Lisp Texts,
Natural Language - Integrated Systems,
AI Tools - VAX LISP Sources,
Theory - Intelligence Tests & Analogy & Common Sense &
Representationalist Perception
----------------------------------------------------------------------
Date: Tue, 8 Jul 86 13:08 EST
From: HAFNER%northeastern.edu@CSNET-RELAY.ARPA
Subject: Lisp texts
Replying to Mark Richer's query about texts for teaching Lisp:
There are a number of good textbooks on Lisp. I prefer Winston & Horn
because of the emphasis on applications of Lisp, especially to AI.
However, whatever text you choose, you should supplement it with
"The Little Lisper" 2nd edition by Dan Friedman and Matthias Felleisen.
TLL is a wonderful teaching tool - it is skill-oriented, thorough,
and entertaining. I expect it will be especially useful for the
students who are not math or CS majors. Good luck!!
Carole Hafner
hafner@northeastern
P.S. Regarding the appropriateness of comments on Lisp programming
on the AILIST: I find this material interesting, relevant, and highly
appropriate. Lisp is the medium for most AI research, and effective
use of that medium is of great concern to many. Ditto for other programming
methods (logic programming, object oriented programming, etc.)
------------------------------
Date: Tue, 8 Jul 86 09:09:13 edt
From: Eric Nyberg <ehn0%gte-labs.csnet@CSNET-RELAY.ARPA>
Subject: Re: Architectures for interactive systems?
> There seems to have been a great deal of work done in
> natural language processing, yet so far I am unaware of
> any attempt to build a practical yet theoretically well-
> founded interactive system or an architecture for one.
...
> Many of the sub-problems have been studied at least once.
> Work has been done on various types of necessary response
> behavior, such as clarification and misconception correction.
> Work has been done on parsing, semantic interpretation, and
> text generation, and other problems as well. But has any
> work been done on putting all these ideas together in a
> "real" system?
...
> I would be interested to get any references to work on such
> integrated systems. Also, what are people's opinions on this
> subject: are practical NLP too hard to build now?
> Brant Cheikes
I am part of a research project that has been investigating
integrated architectures for intelligent interfaces at
GTE Laboratories. A good overview of our recent work can be
found in the Summer issue of IEEE Expert, in a paper entitled
"An Intelligent Database Assistant" [Jakobson 86].
The phrase "practical yet theoretically well-founded" strikes
at one of the basic difficulties in building a natural language
interface as part of a working system - it should work in a
reasonable amount of time, yet be as flexible as possible in
the different kinds of utterances it can understand. The two
extremes are seen in a keyword-based system, where parsing is done by
a hand-coded program, versus a formally complete English grammar system,
where parsing is done by a large, complex data structure (e.g., an ATN).
The simplifying requirement we have placed on our applications is
quite similar to the desire for a narrow, well-defined domain
in building expert systems. If the domain of application for the
intelligent interface is well-defined, and fairly narrow,
a semantic grammar approach can be used quite successfully to
provide good performance with reasonably complete coverage.
The semantic grammar approach that we use is based on case
theory, a linguistic paradigm that was investigated in the
late sixties and early seventies (for a good summary of case-
based approaches, see [Bruce 75]). The case-frame approach to
parsing natural language has also been researched by Jaime
Carbonell, Phil Hayes [Hayes 85], and others at CMU. Case frame
parsing forms the basis for the Language Craft product offered
by Carnegie Group.
Of course, there are some drawbacks to the approach, most notably
a somewhat informal, arbitrary definition of syntax, which makes
the case-frame approach less satisfying from a theoretical
linguistic viewpoint. However, some of the more complex syntactic
constructions (like relative clauses) seem to be less important in
this kind of system than discourse phenomena like ellipsis and
anaphora. The dialog our system has with a user is very
task oriented, and generally does not require the more complex
constructions of unrestricted English prose.
In my opinion, "practical" and "theoretically well-founded" are two
qualities that a natural language system can have, and for each
potential application, the proper mix of efficiency and coverage
must be found.
-- Eric Nyberg
References
----------
[Bruce 75]
Bruce, B., "Case Systems for Natural Language," Artificial
Intelligence, Vol. 6, No. 4, April 1975, pp. 327-360.
[Hayes 85]
Hayes, P., et. al., "Semantic Caseframe Parsing and Syntactic
Generality," Proc. 23rd ACL, Jul. 1985, pp. 153-160.
[Jakobson 86]
Jakobson, G., et. al., "An Intelligent Database Assistant,"
IEEE Expert, Vol. 1, No. 2, Summer 1986, pp. 65-78.
{other references to intelligent interfaces can be found in the
bibliography of [Jakobson 86]}
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
CSNET: ehn0@gte-labs Eric H. Nyberg, 3rd
UUCP: ..harvard!bunny!ehn0 GTE Laboratories, Dept. 317
ARPA: ehn0%gte-labs@csnet-relay 40 Sylvan Rd.
Waltham, MA 02254
(617) 466-2518
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
------------------------------
Date: Wed, 9 Jul 86 16:55:44 pdt
From: saber!matt@SUN.COM (Matt Perez)
Subject: Re: Architectures for interactive systems?
> I would be interested to get any references to work on such
> integrated systems.
Sorry, I have only a vague reference (see below), but I
do have a couple of comments.
> Also, what are people's opinions on this
> subject: are practical NLP too hard to build now?
I don't think it is impossible to integrate such a system.
For example, the *Unix Consultant* at UCB is such an
integrated system, albeit for research rather than
commercial purposes. But its application is practical
enough: to provide an on-line Unix expert which can
communicate with the user in natural language, for
input as well as in its responses.
> Should we
> leave the construction of practical systems to private enter-
> prise and restrict ourselves to the basic research problems?
Lord, NOOOOOOOOOOOO. The integration work is just
beginning and I suspect it is a lot more complicated than
taking care of the individual subproblems. I'd say that
"the construction of practical systems" IS a basic
research problem. All that private enterprise can
afford to do is implement what works, and as you well
pointed out, ain't much that works so far.
As an alternative, I offer that Natural Language
by itself is not that natural a way to communicate
anyways. In many instances a Graphical Interface is
much more appropriate. Of course, by Natural Language I
mean written language or even speech; by Graphical
Interface I mean Graphics (generative and otherwise)
display and feedback and input devices that exploit our
kinetic abilities. Thus I rather point at a feature in
a good display than describe the same feature verbally.
If you don't agree with me on that, try to describe a
scene to someone over the phone.
In other instances, formulae is the communications tool
of excellence. It depends. Ideally, I say, the user
interface should support all of the above!
Basically, however, I agree with you in the following
way: let's first learn to build systems (and enumerate
architectures) that support (solely) a Natural Language
interface. Ditto for graphics. Ditto for formulae.
Then, let's see if we can take the best of each and put
them together reliably and appropriately. And if that
ain't basic research ...
* Matt Perez * DISCLAIMER: beis-ball has bean bery, bery guud too me
matt@saber.uucp sun!saber!matt@decwrl.dec.com ...{ihnp4,sun}!saber!matt
Saber Technology Corp / 2381 Bering Drive / San Jose, CA 95131 (480) 435-8600
------------------------------
Date: Mon, 7 Jul 86 22:51:19 edt
From: beer%case.csnet@CSNET-RELAY.ARPA
Subject: VAX LISP Sources
In a previous AIList (Vol. 4, Issue 127), I posted a message concerning
the availability of a set of tools and utilities for VAX LISP. At that
time, only the object code was in the public domain. However, by
popular request, we have arranged to make the source code for these
facilities available. Anyone who requested a tape of the object code will
be sent the source. The description of the facilities is repeated below.
Here at the Center for Automation and Intelligent Systems Research at
Case Western Reserve University, we have developed a number of tools and
utilities for VAX LISP. They include extensions to the control and string
manipulation primitives, a simple pattern matcher, a pattern-based APROPOS
facility, a pattern-based top-level history mechanism, an extensible top-level
command facility, an extensible DESCRIBE facility, and an implementation of
Flavors. These facilities are described in detail in a technical report,
"CAISR VAX LISP Tools and Utilities" (TR-106-86).
A tape containing the VAX LISP source for these facilities is available for
a $35 shipping and handling fee.
Randall D. Beer
(beer%case@CSNet-Relay.ARPA)
Center for Automation and Intelligent Systems Research
Case Western Reserve University
Glennan Bldg., Room 312
Cleveland, OH 44106
------------------------------
Date: 7 Jul 1986 1059-PDT (Monday)
From: Eugene miya <eugene@ames-aurora.arpa>
Subject: A comment to an interesting posting to net.ai
<"Expert systems" are not AI.>
The following appeared on the USENET's net.ai list (distinct from
the mod.ai list gateway to the ARPAnet. My commentary follows:
>From: michaelm@bcsaic.UUCP (michael maxwell)
>Subject: Re: The Turing-Ring Test -- A Limitation Game.
>Message-ID: <589@bcsaic.UUCP>
>Date: 3 Jul 86 17:14:26 GMT
>
>In article <7101.8606281319@maths.qmc.ac.uk> gcj@qmc-ori.UUCP (The Joka):
>>The following test has been proposed. Appoint one (or more)
>>adjudicators to decide on which of the two parties in the
>>test, persons A and B, is talking to a telephone answering
>>machine and which is talking to a human being. This test is
>>not limited to textual information, although person A should
>>relay the same information as person B.
>
>Wonderful idea! An even better idea: You've probably answered the phone,
>only to find that the voice on the other end is a computerized "survey". I
>propose the following test: which of two computerized "survey"
>programs is talking to a telephone answering machine and which is talking to a
>human being...:-)
>--
>Mike Maxwell
>Boeing Artificial Intelligence Center
> ...uw-beaver!uw-june!bcsaic!michaelm
I have been thinking about the characteristics of a real Turing test.
Here are some thoughts and some questions. 1) The Turing test is basically
a psychological test of "discrimination" [a loaded word in our society
today]. 2) given that the task is to create a machine "with intelligence,"
a) how long should such a test be? b) what is the shortest `length'
of such a test? 3) Since the objective is whether a machine is
intelligent or not (as opposed to `how' intelligent, i.e. an `intelligence
test'), how should the test be composed? It seems that it can be made a
signal detection task, and if so, it will have the standard concepts
of false-positives and true-negatives (all that stuff from radar).
It seems that such a test would be composed of rather difficult questions
of the type: "Your wife (husband) and your daughter (son)have fallen into
the water. You are positioned in the middle and can only save one.
Who do you save?"
Single difficult questions are probably insufficient. Are aggregate
questions any better? Humans are bound to `fail' many questions.
Such questions would be great for a conference to be held in say 2000
when the 50th anniversery of Turing's original paper was published.
>From the Rock of Ages Home for Retired Hackers:
--eugene miya
NASA Ames Research Center
eugene@ames-aurora.ARPA
"You trust the `reply' command with all those different mailers out there?"
{hplabs,hao,dual,ihnp4,decwrl,allegra,tektronix,menlo70}!ames!aurora!eugene
[Sounds like the test for androids in Blade Runner. The problem of
discriminating between two classes of intelligence is much easier than
that of discriminating intelligence from all possible forms of
nonintelligence. By the way, the fastest way to identify human
intelligence may be to look for questions that a human will recognize
as nonsense or outside his expected sphere of knowledge ("How long
would you broil a 1-pound docket?" "Is the Des Moines courthouse taller
or shorter than the Wichita city hall?") but that an imitator might try
to bluff through. -- KIL]
------------------------------
Date: Tue, 8 Jul 86 17:47:02 edt
From: Jay Weber <jay@rochester.arpa>
Reply-to: jay@rochester.UUCP (Jay Weber)
Subject: Transitivity of a *particular* analogy, and let's do science!
In repsonse to my claim that particular analogies are transitive,
Uttam Mukhopadhyay writes:
>However, the SPACE of analogies does not partition the space of
>categories because the world can concurrently be modeled by multiple
>abstraction lattices (not necessarily hierarchies) in which the
>transitivity property may not hold. Consider the following:
>
> a) "A battery is like a reservoir" (storage capability)
> AND b) "A reservoir is like a pond" (body of water)
>
>DO NOT IMPLY:
> c) "A battery is like a pond"
But I orginally wrote:
>> 3) transitive. "A battery is like a reservoir" and
>> "A reservoir is like a ketchup bottle" imply
>> "A battery is like a ketchup bottle" WHEN THE SAME
>> ANALOGY HOLDS BETWEEN THEM (same R).
Note the use of "SAME ANALOGY" which is not the same as "any analogy"
as is the basis of Uttam's example above. Of course, any two categories
are analogous with respect to some mapping function, so the relation
"is analogous to" is vacuous. This distinction tends to be obscured
by the fact that most linguistic examples of analogy (like those above)
leave the mapping function implicit.
Furthermore, I did not claim that the SPACE of analogies partitions the
space of categories, but that a particular analogy does:
>> Then any analogy R is an equivalence relation, partitioning the space
>> of categories.
I also questioned the value of asking whether "creativity" is equivalent
to "making interesting analogies" to which Uttam replied:
> I am glad that scientists, by and large, have not let "slipperiness" in
>some linguistic sense (as you define it) discourage them from carrying on
>their research.
Proper scientists (by definition) do not construct theories about things
that cannot be empirically examined, e.g. using structure mapping functions
to model the communal descriptive definition of the English word
"creativity". Scientists pick testable domains such as problem solving
where you can test predictions of a particular theory with respect to
correct problem solving. In the past, scientists have left debate over
such concepts as "truth" and "beauty" to philosophers, and I think we
should do the same with "creativity" and "intelligence". In Cognitive
Science, researchers have too often exaggerated the impact of their work
through the careless and unscientific use of such terms.
Jay Weber
Computer Science Department
University of Rochester
Rochester, NY 14627
jay@rochester
------------------------------
Date: 8 Jul 86 17:30 PDT
From: Newman.pasa@Xerox.COM
Subject: Re: Common Sense
Philosophically, Sherry Marcus' ideas about common sense are poor in the
same sense that I think Searle and Dreyfus' ideas about why AI won't
ever happen are poor. As near as I can tell all three end up with some
feature of human intelligence which cannot be automated for basically
unexplained reasons. Marcus' problem is simpler than the others (why
can't a computer have a real world common sense database like a
human's?), but it is the same basic philosophical trap. All three appear
to believe that there is some magical property of human intelligence
(Searle and Dreyfus appear to believe that there is something special
about the biological nature of human intelligence) which cannot be
automated, but none can come up with a reason for why this is so.
Comments?? I would particularly like to hear what you think Searle or
Dreyfus would say to this.
>>Dave
------------------------------
Date: Wed, 9 Jul 86 09:18:47 -0200
From: Eyal mozes <eyal%wisdom.bitnet@WISCVM.ARPA>
Subject: Re: Representationalist Perception
> The
> 'representatonalist' account of perception does NOT claim that instead of
> perceiving the world, we perceive internal representations of the world.
> That would indeed be a position with many difficulties. Rather, it says
> that the WAY we perceive the world is BY making representations of it.
> The data structures are, to put it simply, the output of the perceptual
> process, not its input.
I would agree with Gibson (and with Kelley) that this boils down to the
same thing.
The "output" of perception (if such a term is appropriate) is our
awareness. Realists claim that this awareness is directly of external
objects. Representationalists, on the other hand, claim that we are
directly aware only of internal representations, created by a process
whose input are external objects; this means that we are aware of
external objects only INDIRECTLY. That is the position Gibson and
Kelley argue against, and I think they do understand it accurately.
Note that the above applies only to PERCEPTUAL representationalists.
It does not apply to COGNITIVE representationalists, who may agree that
perception is direct, but claim that internal representations are then
formed for the purpose of conceptual thinking. Gibson claimed that
concept-formation is direct as well; but on this point, Kelley
disagrees with him (this is indicated by his discussion of the issue in
chapter 7 of "The Evidence of the Senses"; by his paper "A Theory of
Abstraction", published in "Cognition and Brain Theory", vol. 7, no. 3
and 4, Summer/Fall 1984; and by his references to Ayn Rand's
"Introduction to Objectivist Epistemology").
Eyal Mozes
BITNET: eyal@wisdom
CSNET and ARPA: eyal%wisdom.bitnet@wiscvm.ARPA
UUCP: ..!ucbvax!eyal%wisdom.bitnet
------------------------------
End of AIList Digest
********************
