Copy Link
Add to Bookmark
Report
NL-KR Digest Volume 05 No. 07
NL-KR Digest (8/09/88 22:19:12) Volume 5 Number 7
Today's Topics:
English grammar
Information needed:
Seeking Information on Natural Language Rule Encoders
RE: Sapir/Whorf and fair debate
why Whorf-Sapir is controversial
Re: Sapir-Whorf Hypothesis
Last word on Sapir-Whorf
Call for Papers - AI and Law Conference
Submissions: NL-KR@CS.ROCHESTER.EDU
Requests, policy: NL-KR-REQUEST@CS.ROCHESTER.EDU
----------------------------------------------------------------------
Date: Sun, 31 Jul 88 20:16 EDT
From: John B. Nagle <jbn@glacier.stanford.edu>
Subject: English grammar
I understand that there is an approach to English grammar based on
the following assumptions.
1. There are four main categories of words, essentially nouns,
verbs, adjectives, and adverbs. These categories are
extensible; new words can be added.
2. There are about 125 "special" words, not in one of the four
main categories. This list is essentially fixed. (New
nouns appear all the time, but new conjunctions and articles
never.)
Does anyone have a reference to this, one that lists all the "special"
words?
John Nagle
------------------------------
Date: Fri, 5 Aug 88 08:20 EDT
From: Pierre Guenzburger <mcvax!hslrswi!pierre@uunet.UU.NET>
Subject: Information needed:
I'm posting this for a friend who has not (yet) access to the net.
You can reply to me (pierre@hslrswi.uucp) and I will forward your
message or contact Simon at the address given below.
Simon Moser
Beitenwil
CH-3113 RUBIGEN
Switzerland
+41-31-832692
__________________________________________________________________
For my diploma thesis that I am writing at the University of Berne
I'm looking for related work done elsewhere. The title of the thesis
is 'An Expert System for Database Consistency Rules' (part of a Know-
ledge Based Database Development System). The topics I am interested
in are the following:
1. Expert System Shell in C++
--------------------------
eventually using the RETE-match algorithm
allowing forward and backward chaining
2. Natural Language Processor in C++
---------------------------------
using the ATN-, WASP- or any other technique
allowing dynamic expansion of the vocabulary and/or grammar
3. Grammars for English, eventually German
---------------------------------------
ATN- and WASP-grammars are preferred but any will be of in-
terest
Thank you for reading this message and for your readiness to share
your knowledge!
Sincerely
Simon Moser
------------------------------
Date: Mon, 8 Aug 88 00:32 EDT
Date: Thu, 4 Aug 88 00:48 EDT
From: ERIC Y.H. TSUI <munnari!aragorn.oz.au!eric@uunet.UU.NET>
Subject: Seeking Information on Natural Language Rule Encoders
Can someone provide pointers to AI techniques/programs that encode decision
rules through a natural language (or close to NL) dialog ? (However, I am
not interested in systems that encode a rule atom/assertion as a string
of characters or a proposition.) Presumably, knowledge captured in these rules
are represented in an intermediate notation/formalism. Appreciate any
relevant direction.
Eric Tsui (CSNET: eric@aragorn.oz)
Division of Computing and Mathematics
Deakin University
Geelong, Victoria 3217
Australia
------------------------------
Date: Thu, 4 Aug 88 16:40 EDT
From: HEARNE@wwu.edu
Subject: RE: Sapir/Whorf and fair debate
This is a comment on the tenor of G Fitch's recent controbution
to the Sapir/Whorf debate, to wit:
> Was there any conclusion to this discussion, or did it
> just sort of peter out? I was awaiting some definitive
> word. If there wasn't any, of course I'll provide one,
> but I was hoping someone who knew what they were talking
> about would spare me the trouble.
People who write this sort of thing should be reminded of how
much it demeans them in the perceptions of other subscribers.
We all know that in participating in this sort of communication
we will be exposed to views that seem to us uninformed. If we
couple that knowledge with the normal adult's self-control, we
try to confine our controbutions to ones which advance the
discussion. Insulting _every_ controbutor to a particular
exchange _obviously_ helps no one.
of us
------------------------------
Date: Thu, 4 Aug 88 15:19 EDT
From: Bruce E. Nevin <bnevin@cch.bbn.com>
Subject: why Whorf-Sapir is controversial
The Whorf-Sapir Hypothesis is controversial for reasons that have deep
roots in the opposition of Realist and Idealist perspectives in
philosophy, and in the contradiction of ideologies of the left and right
in personality types.
Idealist/Realist
Naive Realism says things are just as they appear, that our perceptions
correspond reliably to reality. Philosophically credible forms of
Realism back away from this in various ways, e.g. to allow for the
existence of unobservables in physics, but the basic predilection is
there: a WYSIWYG world. This view is typically (maybe necessarily)
materialist: reality consists of matter. Knowledge is justified true
belief.
For Idealism, reality consists of what matters: what objects mean to
us, the assigners of meaning. For an Idealist like Kant, the thing in
itself (Ding an sich) is something essentially and necessarily
unknowable. Things are black boxes whose contents may be inferred but
never directly inspected and known. Knowledge consists of agreements
between observers as to their ways of accounting for the inputs and
outputs of black boxes, agreements that are justified to the extent that
they are supported by objects' definitions of themselves by their inputs
and outputs. (What counts as inputs and outputs is in part a matter of
one's point of view: some differences may be imperceptible, some may be
irrelevant, from a given point of view, but not from another.)
Left/Right
Sylvan Tomkins some 25 years ago talked usefully about the value of the
center between extremes of ideology ('Left and Right: A Basic Dimension
of Ideology', in _The Study of Lives: Essays on Personality in Honour
of Henry A. Murray_, 1963; Hampden-Turner, _Radical Man_, 1971. There's
a precis in Hampden-Turner's 1982 book _Maps of the Mind_.) Ideology
consists of a `love affair beteen a loosely organized set of feelings
and a highly organized and articulate set of ideas'. Ideology precedes
any dealing with problems, and predetermines how people frame problem
and solutions.
For the Right, standards of conduct are external to the individual. One
is obliged to live up to them, and if one fails they to be imposed on
one. Integrity is a matter of principle. Ambiguity is frightening
because it reveals that the presumed unambiguous truth about The Way
Things Are is not known. The world is a dangerous place in which more
or less fit individuals struggle for survival, and things unknown are
hazards. Concepts must correspond with reality and with facts,
constrained by careful measurement, exact methods, and rigorous logic.
For the Left, standards of conduct are rooted within the person
as an expression of personal integrity (being integrated as a person).
Ambiguity accompanies every creative opportunity. Cooperation (as
Darwin himself argued) is as important as competition, and the unit of
survival is the social group, not the individual. Poincare' argues that
mathematics is the finest type of human play, a personal delight
constructed by the imagination. Einstein says that `the formulation of
a problem is often more essential than its solution, which is merely a
matter of mathematics and experimental skill . . . Physical concepts are
free creations of the human mind and are not, however it may seem,
uniquely determined by the external world.'
I believe the determinants of one's philosophical perspective are to be
found in this ideological dimension of personality. (Does this
correlate with the `neats' vs the `scruffies'?)
Physical/Psychological/Social Reality
Emil Durkheim introduced the notion of `social facts', as distinct from
psychological and physical facts. The fact that I do not feel free to
remove all my clothing as I walk out to my car to commute home is a
social fact, a definite pressure that I feel despite the physical fact
that it is nearly 100 degrees out and very humid, and the psychological
fact that I wish I could and would feel much more comfortable if I did.
The means by which I would be constrained if I did this involve all
three kinds of facts. (I do respond differently to a person in a police
uniform than to other physical objects, don't you?)
When you are discussing bricks and billiard balls, quarks and muons,
Realism seems right (though there is some question about quantum effects
with the muons--Bell's theorem and all that.)
When you are discussing gestures and body language, politics and
conversation, Realism amounts to bigoted ethnocentrism. "He nodded his
head yes!" No, he's Greek, and that lifting nod with a barely audible
"tsk" means "no".
From an Idealist perspective, let's say we can get away with Realism
when we're talking about loaves of bread (much as we can get away with
Newtonian physics for engineering purposes). However, we cannot get
away with simple Realism when we are talking about the purchase of
loaves of bread and about eating habits, because we are talking
significance, and that gets us into cultural relativity.
Much of the knowledge of interest and value to be represented in KR
and AI systems is knowledge of social facts. Certainly, where users are
involved, a system must be sensitive to social facts to be judged
intelligent. Saying that there is no record for AI202 is an accurate
statement about physical facts in the database. Presenting this to the
user as the answer "None" to the question "How many students are failing
AI202?" is unresponsive precisely because it confuses the two types of
facts, and pretends to knowledge of a social fact that the system in
fact lacks. The programming required to come up with the answer "There
is no AI202 this quarter. Do you mean AI203?" is a beginning at
representing some of the relevant social facts. A better job might be
done if we recognize that this is what we are about.
The Idealist views in Winograd & Flores might be more palatable to those
who now reject them out of hand if they were presented as complementing
Realist ideas, rather than proposing to supplant them. They have to do
with social facts and psychological facts. Realist models are
appropriate for physical facts. Computer systems need to cope
intelligently with more than just physical facts.
Bruce Nevin
bn@cch.bbn.com
<usual_disclaimer>
------------------------------
Date: Fri, 5 Aug 88 04:01 EDT
From: Kennita Watson <kwatson@oracle>
Subject: Re: Sapir-Whorf Hypothesis
I think the speaking/thinking causality is backwards. People will
figure out a way to talk about various types of snow (or whatever) when
they need to point out the difference between various types of snow, and
not before. People make up words all the time. And it's people who had
already noticed the distinctions between various styles of painting who
made up the words "Cubist", "Impressionist", and so on.
As far as its relation to human behavior, I think a lot of it comes down
to laziness. If I need to refer to "that groove that appears when I
curl my toes, which I use to pick up pencils" often enough apart from
the toes themselves, I will think of a word for it. But until it
becomes a problem, I won't bother.
Kennita
-------------------------------------------------------------
Kennita Watson oracle!kwatson@hplabs.hp.com
Oracle Corporation ...hplabs!oracle!kwatson
------------------------------
Date: Fri, 5 Aug 88 07:50 EDT
From: Dr. Robert Frederking <ref@ztivax.siemens.com>
Subject: Last word on Sapir-Whorf
I first ran into this as an undergraduate, taking an Anthropology
minor, around 1975. At the time, the anthropology professor commented that
it was typical of inter-disciplinary work that field X would get excited
about a theory in field Y about the same time that field Y decided it was
definitely false or vacuous. At the time, anthopologists were just
discovering that linguists considered the S-W hypothesis untrue in its
strong form and uninteresting in its weak form.
Remember, this was 1975! It's a wonderfully intriguing hypothesis
that everyone is just sure is true when they first hear it, but which really
doesn't amount to much. Yes, if you have only one word for snow, you'll
think of all snow as the same. But if you have a reason to distinguish
different kinds, you'll immediately form descriptive phrases for each, and
if the distinction remains important, you'll give them each names. You
probably can figure out what is important in a culture partly by looking at
its language, but neither "controls" the other in any strong way. There are
over 200 words in English for "nails", but I can't tell a brad from a carpet
tack from a masonry nail. The introduction of the tomato revolutionized
Italian cooking, and added one word to the language. And so on.
------------------------------
Date: Fri, 5 Aug 88 14:12 EDT
From: carole hafner <hafner@corwin.ccs.northeastern.edu>
Subject: Call for Papers - AI and Law Conference
CALL FOR PAPERS
Second International Conference on
ARTIFICIAL INTELLIGENCE and LAW
June 13-16, 1989
University of British Columbia
Vancouver, British Columbia, Canada
The field of AI and Law -- which seeks both to understand fundamental mechanisms
of legal reasoning as well as to develop useful applications of AI to law --
is burgeoning with accomplishments in both basic research and practical
applications. This increased activity is due in part to more widely available
AI technology, advances in fundamental techniques in AI and increased interest
in the law as an ideal domain for studying certain issues central to AI.
The activities range from development of classic expert systems, intended as
aids to lawyers and judges, to investigation of canonical elements of case-based
and analogical reasoning. The study of AI and law both draws on and contributes
to progress in basic concerns in AI, such as representation of common sense
knowledge, example-based learning, explanation, and non-monotonic reasoning,
and in jurisprudence, such as the nature of legal rules and the doctrine
of precedent.
The Second International Conference on Artificial Intelligence and
Law (ICAIL-89) seeks to stimulate further collaboration between workers in
both disciplines, provide a forum for sharing information at the cutting
edge of research and applications, spur further research on fundamental
problems in both the law and AI, and provide a continuing focus for the
emerging AI and law community.
Authors are invited to contribute papers on topics such as the following:
-- Legal Expert Systems
-- Conceptual Information Retrieval
-- Case-Based Reasoning
-- Analogical Reasoning
-- Representation of Legal Knowledge
-- Computational Models of Legal Reasoning
In addition, papers on relevant theoretical issues in AI (e.g., concept
acquisition, mixed paradigm systems using rules and cases) and in
jurisprudence/legal philosophy (e.g., open-textured predicates, reasoning
with precedents and rules) are also invited provided that the relationship
to both AI and Law is clearly demonstrated. It is important that all authors
identify the original contributions presented in their papers, exhibit
understanding of relevant past work, discuss the limitations as well as
the promise of their ideas, and demonstrate that the ideas have matured
beyond the proposal stage. Each submission will be reviewed by at least three
members of the Program Committee and judged as to its originality, quality,
and significance.
Authors should submit six (6) copies of an Extended Abstract, which must include
a full list of references, by January 10, 1989 to the Program Chair:
Edwina L. Rissland
Department of Computer and Information Science
University of Massachusetts, Amherst, MA 01003, USA;
(413) 545-0332, rissland@cs.umass.edu.
Submissions should be 6 to 8 pages in length, not including references.
No electronic submissions can be accepted. Notification of acceptance or
rejection will be sent out by early March. Final camera-ready copy of the
complete paper (up to 15 pages) will be due by April 15, 1989.
Program Chair: Edwina L. Rissland, University of Massachusetts/Amherst and
Harvard Law School
General Co-Chairs: Robert T. Franson, Joseph C. Smith, Faculty of Law,
University of British Columbia
Secretary-Treasurer: Carole D. Hafner, Northeastern University
Program Kevin D. Ashley IBM Thomas J. Watson Reasearch Center
Committee: Trevor J.M. Bench-Capon University of Liverpool
Donald H. Berman Northeastern University
Jon Bing University of Oslo
Michael G. Dyer UCLA
Anne v.d.L. Gardner Palo Alto, California
L. Thorne McCarty Rutgers University
Marek J. Sergot Imperial College London
------------------------------
End of NL-KR Digest
*******************
