AIList-REQUEST@SRI-AI.ARPA (AIList Moderator Kenneth Laws) (12/18/85)
AIList Digest Wednesday, 18 Dec 1985 Volume 3 : Issue 189
Today's Topics:
Query - Finding Loops in Prolog & CAI,
AI Tools - Object oriented programming in Common Lisp,
Policy - ADS Message & Advertisements,
Logic - Counterfactuals
----------------------------------------------------------------------
Date: Tue, 17 Dec 85 15:34:54 est
From: Catherine A. Meadows <meadows@nrl-css.ARPA>
Subject: loops
Does anyone out there know of any work that has been done on
checking for loops in Prolog (other than the papers recently published
in ACM Sigplan)?
Cathy Meadows
meadows@nrl-css
------------------------------
Date: 17 Dec 85 23:38 EST
From: Gunther @ DCA-EMS
Subject: REQUEST FOR INFO ON CAI ORIENTED SHELLS
IF ANYONE IS WORKING IN THE AREA OF EXPERT CAI DEVELOPMENT
SYSTEMS, I'D LIKE SOME INFORMATION ON APPROACHES, OTHER WORK,
PACKAGES, ETC ...
THANKS,
J.L. FEINSTEIN
[Contact AI-Ed-Request@SUMEX-AIM.ARPA to get in touch with
a like-minded group. -- KIL]
------------------------------
Date: 17 DEC 85 13:32-N
From: DESMEDT%HNYKUN52.BITNET@WISCVM.WISC.EDU
Subject: Object oriented programming in Common Lisp
In reply to the following request:
From: Nick Davies (at GEC Research) <YE85%mrca.co.uk@ucl-cs.arpa>
Subject: Object oriented programming in Common Lisp
> Does anyone have or know of an implementation of Flavors or any other
> object-oriented programming system in Common Lisp ?
I would like to mention CORBIT, which is ORBIT rewritten in NIL Common
Lisp at the University of Nijmegen (the Netherlands).
CORBIT is an object oriented extension of Common Lisp which is NOT based
on message-passing but on the idea of generic functions. An ultra-short
example illustrating the difference:
Flavors: (send window-1 :expose)
CORBIT: (expose window-1)
What is seen as a message in Flavors is a function in CORBIT. So it is
not the object, but the message which is functional. This has a number of
advantages. One advantage is that it is straightforward to use ordinary
lambda-binding to localize inherited definitions. Another one is that the
generic functions can be traced like any other Lisp function. The overall
consequence is that CORBIT fits more neatly in the Lisp way of thinking
than message-passing systems, is simpler to understand and implement, and
yet offers the same possibilities.
Koenraad De Smedt, DESMEDT@HNYKUN52 (bitnet)
Psychological Laboratory
University of Nijmegen
The Netherlands
[The recent Xerox PARC work on CommonLoops has a similar functional
flavor -- as does ADA, of course. -- KIL]
------------------------------
Date: Mon 16 Dec 85 14:00:25-PST
From: Wilkins <WILKINS@SRI-AI.ARPA>
Subject: AIDS
Despite Duffy's bafflement, it seems obvious why they changed
their name and it is important to the community to learn of
such name changes. But, isn't there already someone named
Advanced Decision Systems? Sounds familiar, but I cannot recall . . .
------------------------------
Date: Tue, 17 Dec 85 17:40:04 EST
From: David_West%UMich-MTS.Mailnet@MIT-MULTICS.ARPA
Subject: Policy
You said in AIList v3 #185:
"I [...] do screen out job ads [...]"
For those of us out here in the boonies, the nets are the only available
*current* source of information on what's really happening: the print media
are agonizingly slow, particularly if one must rely on a university library.
From time to time, we boonie-dwellers even need to find jobs, and most AI
jobs are outside the boonies.
You didn't say whether you screen out job ads because of ARPAnet policy,
or because they have a better home elsewhere. If it's the former, I think
the policy is misguided, at least for academic/research jobs. If it's the
latter, and there is in fact a network source for this kind of information,
I don't know of it, and would like to.
(I have seen *very* occasional academic fellowship announcements here:
perhaps as many as two in the last year. Presumably there were more that
went unreported.)
[There are many reasons for the policy. (1) There are the Arpanet
restrictions on commercial use: academic ads are generally considered
acceptable but industrial ads are not. (I don't understand the
distinction -- placement of personnel in the defense industry is in
the interest of the net's military sponsors -- but rules are rules.)
Even if Arpanet sponsors and host administrators did not object,
a fair proportion of any such messages draw flak from other readers;
I don't need the hassle of defending my editorial policies that often.
(2) There is already a channel, Arpanet-BBoards@MIT-MC, for academic
announcements and certain commercial announcements; I prefer to
avoid duplicate traffic. (3) AIList carries enough traffic without
adding another class of messages, particularly one that is of little
interest to most readers and of no value in the historical archive.
One difficulty in pruning the list is in screening out duplicates
of messages submitted in previous months or years -- I have to pull
the archives and scan for half-remembered occurrences. (4) I am
unable to draw a clean distinction between messages of AI interest
and those that are not. A search for a lab director or lecturer
is of interest, but how about an ad for students to run an AI lab's
computers? (4) A few of the people submitting ads lack perspective
and are difficult to deal with -- any attempt to reject just some
of the messages or to modify them to make them less commercial
may lead to multiple interchanges with the authors. It is easier
to reject such messages entirely.
If someone else wants to moderate a list of placement ads or of
resumes, he (or she) will probably find it acceptable to the net
community. The volume of such solicitations put out by certain
AI departments indicates a desire for such a service. (IEEE tried to
serve this need among electrical engineers and computer scientists
with their Professional Abstracts Registry; it couldn't be sustained,
but the task has been passed on to a commercial information server.)
I just find it better not to mix this traffic with AIList. I suggest
that you check CACM and IEEE Computer for recent classified ads, or
just contact companies listed in any recent AI magazine or conference
proceedings. There are also some recruiters specializing in AI
(Halbrect Associates, Inc.; JDG Associates, Ltd; Klein/Thaler Executive
Search; Artificial Intelligence Referral Service; S.J. Parker and
Associates); look for their ads in IEEE Spectrum and the AI magazines.
-- KIL]
------------------------------
Date: 17 Dec 85 08:48:00 EDT
From: "CUGINI, JOHN" <cugini@nbs-vms.ARPA>
Reply-to: "CUGINI, JOHN" <cugini@nbs-vms.ARPA>
Subject: counterfactuals
>> From: Mike Dante <DANTE@EDWARDS-2060.ARPA>
>>
>> (0) Suppose a class consists of three people, a 6 ft boy (Tom),
>> a 5 ft girl (Jane), and a 4 ft boy (John). Do you believe the
>> following statements?
>>
>> (1) If the tallest person in the class is a boy, then if the tallest
>> is not Tom, then the tallest will be John.
>> (2) A boy is the tallest person in the class.
>> (3) If the tallest person in the class is not Tom then the tallest
>> person in the class will be John.
>>
>> How many readers believe (1) and (2) imply the truth of (3)?
> It seems to me that this example gives insight as to what the status
> is of a counterfactual whose premise is true. The general view among
> philosophers on this is that the truth or falsity of a counterfactual
> with a true premise depends simply on the truth or falsity of its
> conclusion, but this example seems to run against this view.
> The reason seems to me to be that the conclusion should be investigated
> as if the premise had read, "If the tallest person in the class is
> *necessarily* a boy." In other words, in constructing possible worlds
> in which "the tallest is not Tom" (i.e., in investigating the truth of the
> conclusion of (1)), possible worlds in which the tallest is not a boy are
> disallowed. Thus (1) and (2) can be true while (3) is false.
I don't think there's a problem here. If we read the phrase "if
the tallest is not Tom, then the tallest will be John"
truth-functionally, in BOTH (1) and (3) then it's clearly true,
since the antecedent is false - Tom is the tallest - and hence
(1) and (3) are both true. If we read it counterfactually, eg:
"what would happen if Tom shrank to a height of 3 angstroms,
leaving only John and Jane? Would John be the tallest, true or
false?" then it's clearly false in both cases, and so (1) and
(3) are BOTH false. If we read (1) as:
(1) If the tallest person in the class is a boy, then what would
happen if Tom shrank to a height of 3 angstroms, leaving only
John and Jane? Would John be the tallest, true or false?
Modus ponens remains with its virtue intact, in either case.
It's only when we read the phrase truth-functionally in (1),
and counterfactually in (3) that an apparent conflict arises.
Further confusing the issue is that (1) contains two implications,
and the first sounds very truth-functional, leading you to read
the second implication in the same way; but in isolation, in (3),
it sounds counterfactual. If someone regards this as a real
problem let him/her express it formally, and use different
signs for different implication, eg "=>" for truth-functional,
"->" for counterfactual.
John Cugini <Cugini@NBS-VMS>
------------------------------
Date: 17 Dec 85 16:51:21 EST (Tue)
From: Dana S. Nau <dsn@rochester.arpa>
Subject: Re: interesting counterfactual
From: Matthew Ginsberg <SJG@SU-AI.ARPA>
> From: Mike Dante <DANTE@EDWARDS-2060.ARPA>
> (0) Suppose a class consists of three people ...
... the conclusion should be investigated as if the premise had read,
"If the tallest person in the class is *necessarily* a boy." In other
words, in constructing possible worlds in which "the tallest is not
Tom" (i.e., in investigating the truth of the conclusion of (1)),
possible worlds in which the tallest is not a boy are disallowed. ...
I don't think this solves the problem. Regardless of whether you allow such
worlds or disallow them, you will still be allowing some worlds in which
John is the tallest, and allowing others in which he is not.
I don't know what to make of this generally. It appears that the problem
can only arise with enbedded counterfactuals ...
I disagree. In Dante's example, the APPARENT problem was that the strict
logical interpretation of the statements didn't pin us down to the particular
counterfactual world that we "obviously" wanted. But I think the real
problem is more general. It's analogous to the frame problem: if we're
going to deny some fact in order to create a counterfactual world, then what
OTHER things are to be changed and what things are to remain the same? In
general, it may not be clear which counterfactual world we want.
In Dante's example above, the "obvious" counterfactual world was one in
which Tom did not exist and all the other axioms were unchanged. But one
could easily specify examples in which the removal of Tom would cause
inconsistency unless some of the other axioms were changed too. In cases
such as this, there may be many possible ways to change the axioms. It
shouldn't be too difficult to construct an example in which different kinds
of changes would seem best to different people--and thus different people
would reach conflicting conclusions about what the changed world would be
like.
------------------------------
End of AIList Digest
********************