NL-KR-REQUEST@CS.ROCHESTER.EDU (12/22/87)
NL-KR Digest (12/21/87 20:42:12) Volume 3 Number 64
Today's Topics:
Where is the Linguistic Society of America meeting this year?
Reference to YANLI wanted
Two requests
online dictionary needed
ELIZA ???
semantics of "unless"
Submissions: NL-KR@CS.ROCHESTER.EDU
Requests, policy: NL-KR-REQUEST@CS.ROCHESTER.EDU
----------------------------------------------------------------------
Date: Thu, 17 Dec 87 14:32 EST
From: George Entenman <ge@mcnc.org>
Subject: Where is the Linguistic Society of America meeting this year?
I would like to find out where and when the Linguistic Society
of America (LSA) is meeting this year. Thanks.
George Entenman
The Microelectronics Center of North Carolina
919/942-5858
------------------------------
Date: Thu, 17 Dec 87 14:52 EST
From: rutgers!lehi3b15.csee.lehigh.edu!keithw
Subject: Reference to YANLI wanted
I am posting a querry for a friend on NL. He claims that there is a
package called YANLI, Yet Another Natural Language Interface. It there
indeed such a package and is it PD? Any pointers to this work would
be very useful.
Thanks in advance.
Keith
------------------------------
Date: Thu, 17 Dec 87 21:46 EST
From: Steve Bradtke <BRADTKE%cs.umass.edu@RELAY.CS.NET>
Subject: Two requests
I have two requests for information and suggestions from the
NL-KR readers. I will be happy to post a summary of the results if
there's enough interest.
1) I am very interested in connectionist approaches to
language, particularly language acquisition and language/world
knowledge representation. I'd like to put together a reading
list to use in an independent study course next term. I believe that
I'm already familiar with most of the work that explicitly tries to
relate connectionist ideas to NL problems. Some examples of these
are the works of Cottrell, McClelland, Kawamoto, Rumelhart, and Pollack.
I am also familiar with the informal debate between Lakoff and Pinker
and Prince on connectionist NL.
Could you recommend a set of books or articles that
you think relate (possibly, perhaps peripherally) to connectionist
models of language? I have been thinking of choosing selections
from the bibliography to Lakoff's "Women, Fire, and Dangerous Things",
but I don't trust myself to determine which would be the more
important sources. All suggestions are welcome, whether they be to work
that is nominally from psychology, linguistics, anthroplogy, or whatever.
Please don't feel that you have to restrict yourself to anything thatf
explicitly mentions connectionism, there won't be too much of that.
2) There are only a few systems to date that have applied
connectionist techniques to language. The work done by Rumelhart
and McClelland, and McClelland and Kawamoto is still representative
of the state of the art. All of these systems that have been
concerned with the _meaning_ of words (McClelland and Kawamoto, for
example) have followed the premise that word meanings are actually
compositions of semantic primitives. Assuming this is true, we are
still left with the problem of discovering a canonical set of
semantic primitives, and then using them to define our lexicon.
In pursuit of solutions to this problem, I have become aware
of Robert Amsler's work, and have been told to contact Roy Byrd and
Judy Kegl.
Would you be able to send me a list of references that
would familiarize me with the work that has been done in the
linguistics community on the (more or less) automatic extraction
of semantic structure from dictionaries?
Thank you very much for your time,
Steve Bradtke
University of Massachusetts at Amherst
------------------------------
Date: Fri, 18 Dec 87 11:17 EST
From: rolandi <rolandi@gollum.Columbia.NCR.COM>
Subject: online dictionary needed
I am trying to locate an online dictionary or any large collection
of English words in electromic form that includes a pronunciation
key. Not to be picky, but the pronunciation key would ideally
employ regular ascii characters to represent the word's phonetic
qualities. Does a "shareware" version of such a resource exist?
Thanks in advance...
w.rolandi
u.s.carolina dept. of psychology and linguistics
ncr advanced development
------------------------------
Date: Sun, 20 Dec 87 17:52 EST
From: Wolf-Dieter Batz <L12%DHDURZ1.BITNET@CUNYVM.CUNY.EDU>
Subject: ELIZA ???
Resent-From: Ken Laws <Laws@KL.SRI.COM>
Resent-To: nl-kr@CS.ROCHESTER.EDU
Hello ppl on AIlist,
I'm in need of help from some of you cracks who have a broader perspective
than me, 'cause I did not read this list for several months now. Question:
Is there any form of some intelligent interviewing software out on the net?
We would apprecciate any source in any language you like. Best would be some
Prog combining the features from ELIZA and STORYTELLER. Please send it to my
address directly, 'cause I do not read this list (no time, believe me...)!
If there's any substantial response, I will post it as a large package to
the list next spring, ok?
thanxalot *** Wodibatz (L12@DHDURZ1.bitnet)
------------------------------
Date: Wed, 16 Dec 87 08:28 EST
From: Michel Lang <lang@bigburd.PRC.Unisys.COM>
Subject: semantics of "unless"
Does anybody have any thoughts about the truth-functional meaning
of the connective "unless"? What I'm getting at is the following:
I recently came across a problem that asked me to prove that
P unless Q
and I puzzled for quite some time over what exactly was being asked.
For example, is it one of the following:
P ==> ~Q
P <==> ~Q
I have since decided in favor of P <==> ~Q, but the number
of disagreeing answers I received from colleagues made me think
it might be worth posting. Any ideas?
--
Francois-Michel Lang
Paoli Research Center, Unisys Corporation lang@prc.unisys.com (215) 648-7469
Dept of Comp & Info Science, U of PA lang@cis.upenn.edu (215) 898-9511
------------------------------
Date: Fri, 18 Dec 87 11:12 EST
From: Varol Akman <varol@cwi.nl>
Subject: Re: semantics of "unless"
lang@bigburd.PRC.Unisys.COM (Michel Lang) writes:
>Does anybody have any thoughts about the truth-functional meaning
>of the connective "unless"?
Please forgive this reply if it sounds too obvious or dumb but
methinks
P unless Q
can be understood as (from a logical viewpoint)
P should be implied by ~Q
Let's take an example:
Continue SDI unless Russians leave Afghanistan
should be understood as
Abandon SDI if Russians leave Afghanistan
Am I missing something here?
-Varol Akman
CWI, Amsterdam
------------------------------
Date: Fri, 18 Dec 87 22:34 EST
From: ROTH <chiefdan@vax1.acs.udel.EDU>
Subject: Re: semantics of "unless"
lang@bigburd.PRC.Unisys.COM (Michel Lang) writes:
>> P unless Q
I'd say "P unless Q" means:
P xor (Q and ~P)
The simplification is left to the student as an exercise.
--
ARPA: chiefdan@vax1.acs.udel.edu "Germany in Jan. Back in Feb."
UUCP: ...!udel!udccvax1!chiefdan
"They nailed him to the cross, and they laid him in the ground, but they should
have known you can't keep a good man down." -- Larry Norman
------------------------------
Date: Sat, 19 Dec 87 14:30 EST
From: Adam J. Kucznetsov <adam@cunixc.columbia.edu>
Subject: Re: semantics of "unless"
lang@bigburd.PRC.Unisys.COM (Michel Lang) writes:
> [query about functional meaning of connective 'unless']
>
> [for example, does P unless Q mean ...]
>
> P ==> ~Q
> P <==> ~Q
In article <149@piring.cwi.nl> varol@cwi.nl (Varol Akman) replies:
> P unless Q
>can be understood as (from a logical viewpoint)
> P should be implied by ~Q
>
>Let's take an example:
> Continue SDI unless Russians leave Afghanistan
>should be understood as
> Abandon SDI if Russians leave Afghanistan
>
>Am I missing something here?
(by the way, "abandon SDI if Russians leave Afghanistan" is an
interpretation under P -> ~Q, i.e. Q -> ~P, and not under ~Q -> P)
sure. ~A -> B is vacuously true when ~A is false, that is when A is
true... in particular, A -> B is always true unless :-) A=T, B=F holds.
in your example, "abandon SDI if Russians leave Afghanistan" is indeed
implied by (but not equivalent to) "continue SDI unless Russians leave
Afghanistan." what if the Russians don't leave Afghanistan? then SDI
could with those orders still, consistently, be abandoned -- but not
with the "unless" order.
i don't buy either P->~Q or ~Q->P for 'P unless Q'. the first sentence
says that the only wrong case is if SDI is continued and the Russians
leave Afghanistan, to continue with the example. the second sentence
says that the only wrong case is if SDI is stopped and the Russians stay
in Afghanistan. strictly speaking, both cases are wrong. (however,
consider the difference between temporal english statements and strict
logic, in which symmetry takes precedence over politics.)
it seems pretty clear to me. A unless B means A in all cases where not
B. in a system constrained only by that sentence, the truth table is
A B A unless B
- - -
F F F russians still there, sdi stopped (no good)
F T T russians still there, sdi continued (ok)
T F T russians left, sdi stopped (ok)
T T F russians left, sdi continued (no good)
it will quickly be noticed that (A unless B) is equivalent to ~(A<->B)
which is, of course, equivalent to (A<->~B) as adopted by the poster who
asked. that is the same as (A xor B). one or the other, not both.
"unless" is symmetric, of course, whereas "implies" is antisymmetric.
HOWEVER!!! English usage of the connective, like that of all the other
ones, is not always equivalent to the formal mathematical use (which i
assume was originally meant, since the question concerned the statement
of a mathematical proof.) in particular, "a unless b" is asymmetric,
since causality or temporality is implied. you wouldn't say "russians
leave afghanistan, unless sdi is continued." (although they might
profess to see it that way :-)
ajk(cat)
--
Cat (Adam) J. Kucznetsov adam@cunixc.columbia.edu cat@dasys1.UUCP
Columbia University, NYC UI.ADAM@CU20B.BITNET AJUUS@CUVMA.BITNET
When in danger, when in doubt, run in circles, scream and shout.
------------------------------
Date: Sun, 20 Dec 87 16:17 EST
From: John Chambers <jc@minya.UUCP>
Subject: Re: semantics of "unless"
> >Let's take an example:
> > Continue SDI unless Russians leave Afghanistan
> >should be understood as
> > Abandon SDI if Russians leave Afghanistan
> it seems pretty clear to me. A unless B means A in all cases where not
> B. in a system constrained only by that sentence, the truth table is
>
> A B A unless B
> - - -
> F F F russians still there, sdi stopped (no good)
> F T T russians still there, sdi continued (ok)
> T F T russians left, sdi stopped (ok)
> T T F russians left, sdi continued (no good)
No, this isn't the correct English meaning, as can be seen by continuing the
first sentence:
Continue SDI unless Russians leave Afghanistan, at which time
we will reconsider the status of the program.
In other words, the last entry in the truth table is incorrect; if the
Russians were to leave Afghanistan, the lawyers in the US government
would be quick to point out that there was no promise to stop SDI in
that case; only that it would continue as long as the Russians stayed.
Of course, once you realize this, you also realize that the correct
mathematical operator is the plain old 'or' (for which there is no
unambiguous English word :-).
--
John Chambers <{adelie,ima,maynard,mit-eddie}!minya!{jc,root}> (617/484-6393)
------------------------------
Date: Mon, 21 Dec 87 07:46 EST
From: Rob Bernardo <rob@pbhyf.UUCP>
Subject: Re: semantics of "unless"
In article <433@minya.UUCP> jc@minya.UUCP (John Chambers) writes:
+> >Let's take an example:
+> > Continue SDI unless Russians leave Afghanistan
+> >should be understood as
+> > Abandon SDI if Russians leave Afghanistan
+
+> it seems pretty clear to me. A unless B means A in all cases where not
+> B. in a system constrained only by that sentence, the truth table is
+>
+> A B A unless B
+> - - -
+> F F F russians still there, sdi stopped (no good)
+> F T T russians still there, sdi continued (ok)
+> T F T russians left, sdi stopped (ok)
+> T T F russians left, sdi continued (no good)
+
+No, this isn't the correct English meaning, as can be seen by continuing the
+first sentence:
+ Continue SDI unless Russians leave Afghanistan, at which time
+ we will reconsider the status of the program.
+
+In other words, the last entry in the truth table is incorrect; if the
+Russians were to leave Afghanistan, the lawyers in the US government
+would be quick to point out that there was no promise to stop SDI in
+that case; only that it would continue as long as the Russians stayed.
This would make sense to me if you replaced "unless" with "until".
The "until" has to do with time, which fits in with the "at which time".
I fully agree with the analysis with the truth table in it.
--
Rob Bernardo, San Ramon, CA (415) 823-2417
I'm not a bug, I'm a feature. {pyramid|ihnp4|dual}!ptsfa!rob
------------------------------
End of NL-KR Digest
*******************