Copy Link
Add to Bookmark
Report
NL-KR Digest Volume 06 No. 08
NL-KR Digest (Mon Mar 13 01:28:17 1989) Volume 6 No. 8
Today's Topics:
CSLI Calendar, 9 March, $:19
Cognitive Science: Don Perlis
MBR Workshop, please post as appropriate
Submissions: nl-kr@cs.rpi.edu
Requests, policy: nl-kr-request@cs.rpi.edu
Back issues are available from host archive.cs.rpi.edu [128.213.1.10] in
the files nl-kr/Vxx/Nyy (ie nl-kr/V01/N01 for V1#1), mail requests will
not be promptly satisfied. If you can't reach `cs.rpi.edu' you may want
to use `turing.cs.rpi.edu' instead.
---------------------------------------------------------
To: nl-kr@cs.rpi.edu
Date: Wed, 8 Mar 89 16:33:16 PST
From: emma@csli.Stanford.EDU (Emma Pease)
Subject: CSLI Calendar, 9 March, $:19
C S L I C A L E N D A R O F P U B L I C E V E N T S
_____________________________________________________________________________
9 March 1989 Stanford Vol. 4, No. 19
_____________________________________________________________________________
A weekly publication of The Center for the Study of Language and
Information, Ventura Hall, Stanford University, Stanford, CA 94305
____________
CSLI ACTIVITIES FOR THIS THURSDAY, 9 March 1989
2:15 p.m. CSLI Seminar
Cordura Hall Indexicality and Quantified Modal Logic
Conference Room Harry Deutsch
Illinois State University
Abstract below
3:30 p.m. Tea
Ventura Hall
____________
ANNOUNCEMENT
Please note that the CSLI Seminar originally scheduled for 7 March is
now on 9 March at 2:15.
The CSLI Calendar will not be published on 16 and 23 March.
____________
THIS WEEK'S CSLI SEMINAR
Indexicality and Quantified Modal Logic
Harry Deutsch
Illinois State University
Thursday, 9 March, 2:15
Cordura Conference Room
Relations between recent philosophy of language and the semantics of
modality (possible-worlds semantics) have not been good. I attempt to
mediate the dispute by formulating quantified modal logic (QML) so as
to incorporate some insights of the "new theory of reference" (NTR).
This sheds some new light on both QML and the NTR.
____________
SYMBOLIC SYSTEMS FORUM
Ontology and Computers
Ruben Kleiman
Apple Intelligent Agents Group
Friday, 10 March, 3:15
Room 60:61N
This talk will be about artificial intelligence and knowledge
representation, focusing on how to encode knowledge into a computer.
On one hand, Winograd, Flores, and Putnam have advocated a
phenomenological view that abandons the standard mentalist position.
On the other hand, there are also many people (Hayes, McCarthy,
Dennett, and most AI workers) who keep the mentalist position. Dr.
Kleiman will attempt to reconcile these two philosophical positions.
____________
LINGUISTICS DEPARTMENT COLLOQUIUM
A Union Analysis of Noun Incorporation
Donna Gerdts, SUNY at Buffalo
Friday, 10 March, 3:15
Cordura Conference Room
Noun incorporation (NI) has been widely discussed from many
viewpoints; this paper presents yet another analysis of this
phenomonon, one based on three previously proposed and independently
instantiated relational constructs: union, multiattachment, and
cancellation. Not only does this treatment provide an analysis at no
cost to the grammar, but it allows for a typology of NI constructions
appropriate for the array of data currently attested in natural
languages. First, we take a quick look at five other analyses pointing
out their empirical inadequacies. The strength of the union analysis
is that it avoids the limitations of other treatments while
nevertheless presenting a constrained view of NI constructions.
____________
COMMONSENSE AND NONMONOTONIC REASONING SEMINAR
Relating Default and Autoepistemic Logics
Wiktor Marek and Miroslaw Truszczynski
University of Kentucky
Monday, March 13, 3:15pm
MJH 301
We introduce a classification of nonmonotonic context-dependent
reasonings according to the way context is used in derivations. A
reasoning is symmetric if context is used to derive both positive and
negative information. A reasoning is asymmetric if context is applied
to derive negative information only. In the talk we concentrate on
symmetric and asymmetric reasonings in default and autoepistemic
logics. They give rise to several classes of objects: weak extensions
and extensions in default logic, and expansions and robust expansions
in autoepistemic logic. Our results establish correspondence between
weak extensions and expansions (both notions are related to symmetric
reasonings) and extensions and robust expansions (these notions are
related to asymmetric reasonings). We also find an exact character of
the correspondence between notions based on the parsimony principle:
minimal sets closed under defaults and stable sets with minimal
objective parts. This multilevel correspondence between default and
autoepistemic logics pinpoints the exact character of the equivalence
of their expressive powers.
------------------------------
To: nl-kr@cs.rpi.edu
Date: Thu, 9 Mar 89 09:36:46 EST
From: rapaport@cs.Buffalo.EDU (William J. Rapaport)
Subject: Cognitive Science: Don Perlis
UNIVERSITY AT BUFFALO
STATE UNIVERSITY OF NEW YORK
GRADUATE GROUP IN COGNITIVE SCIENCE
PRESENTS
DONALD R. PERLIS
Department of Computer Science
and Institute for Advanced Computer Studies
University of Maryland
and
Department of Computer Science
University of Rochester
"THE FOUR REFERENCES"
Mind is a device for reasoning, thinking. So, what is thought? Who
needs it? Not bacteria. But more complex behavior requires processing
information `about' the world. What is `aboutness', and what good is
it? The world is too complex to always correctly model it or algorith-
mize responses to it. For bacteria, it seems not to matter; they sur-
vive in sufficient numbers without having to deal with this issue. But
we are not so lucky, or rather we are lucky that we are not so lucky,
since it has forced us to evolve ways to deal with incorrect algorithms,
namely, to postulate error in ourselves and, on detecting it, take
corrective action. We will consider the extent to which aboutness may
be explained in terms of this capacity, and the idea that thought
amounts to the exercise of this capacity. Language supports this
activity, so we will look a bit at linguistics. Finally, there are fas-
cinating philosophic positions and arguments that bear on this whole
enterprise.
Monday, March 20, 1988
Noon - 2 P.M.
(bring a brown-bag lunch)
317 Park Hall, Amherst Campus
There will be an evening discussion at 8:00 P.M.,
at Stuart C. Shapiro's, 112 Parkledge, Snyder.
For further information, contact Bill Rapaport, Department of Computer
Science, 636-3193.
------------------------------
To: nl-kr@cs.rpi.edu
Date: Wed, 8 Mar 89 14:19:51 pst
From: Ethan Scarl <ethan@atc.boeing.com>
Subject: MBR Workshop, please post as appropriate
* CALL FOR PARTICIPATION *
IJCAI-89 Workshop on
MODEL-BASED REASONING
Detroit, MI, USA Sponsored by AAAI Sunday, August 20, 1989
Systematically representing and reasoning about the structure and function of
a system offers significant advantage in design, analysis, monitoring,
diagnosis, and control.
This workshop will provide a forum for defining and discussing issues that
arise from or underly the use of Model-Based Reasoning (MBR). This includes
familiar applications in monitoring and diagnosis, and other areas of strong
interest such as system control, planning, analysis, theory construction,
tutoring, or design/redesign.
The workshop seeks papers that offer new approaches and insights to
fundamental problems associated with MBR, some of which are listed below.
This is not a forum for descriptions of system implementations or
applications, although analyses of experiences with MBR may be of interest.
We encourage papers which address questions such as the following:
* What inference mechanisms take best advantage of explicit structural and
behavioral representations? What are their power and limitations?
* Conversely, what do these inference mechanisms require of a structural
representation or a behavioral simulation? To what class of systems is MBR
now applicable in a straightforward and useful way?
* What factors may have led to failure in the use of MBR, or to its
abandonment?
* Is there a complexity barrier beyond which systems cannot be effectively
modeled? Are there types of complexity that affect MBR differently?
* How does MBR apply to qualitative, quantitative, or mixed domain
descriptions? Do MBR's methods apply uniformly to analog and discrete
objects?
* Is there a taxonomy of uncertainty or incompleteness in models? Do any of
these contraindicate the use of MBR?
* Can MBR use arbitrary sources of heuristic knowledge without compromising
its power?
* How can models cope when the domain's behavior exhausts computational
resources? Can models use speed/accuracy tradeoffs?
* Can one infer changed or unknown structure from observed or desired
behavior?
* Are there systematic approaches to the "model construction bottleneck?"
A number of problems of domain representation still challenge the generality
of MBR. For example, how shall our models handle:
~ embedded software;
~ feedback, state-dependency, or other forms of time-dependency;
~ open systems, including sensing of external environments and initial
ignorance of structure;
~ systems with virtually no physical structure (e.g., chemical process
control);
~ globally complex systems, rich in interaction paths between relatively
simple components;
~ systems with complex geometric relationships; or
~ translation between physical and computational models (e.g., pipes and
valves to pressures and flows), including derivation of global
parameters from aggregated models?
To facilitate discussion, acceptance will be limited to 35 papers and 10-12
presentations. There will be an informal proceedings of all accepted
submissions. More formal publication will be pursued if there is sufficient
interest. All attendees will be invited to prepare single page descriptions
of their work for distribution prior to the workshop. Each presented paper
will have a commentator.
The workshop's organizing committee consists of: Randy Davis, MIT; Johan de
Kleer, Xerox PARC; Richard Doyle, JPL; Dan Dvorak, UTexas; Mike Fehling,
Rockwell, Mike Genesereth, Stanford; Walter Hamscher, Price-Waterhouse; Ethan
Scarl (chair), Boeing Computer Services; Narinder Singh, Stanford; and Brian
Williams, Xerox PARC.
The submission format will be short, self-contained papers of no more than
2000 words. Recent extensions and condensations of work already submitted to
IJCAI or elsewhere is permissible, so long as any duplication is explicitly
noted. The preferred medium for submission is electronic mail of plain ASCII
files without formatting commands or control characters, and sent to:
ethan@atc.boeing.com
Figures may be faxed to (206) 865-2946, unless their quality or quantity is
prohibitive. Alternatively, three hardcopies may be sent to:
Ethan Scarl Boeing Computer Services, P.O. Box 24346, M/S 7L-64 Seattle,
Washington, 98124-0346
(206) 865-3255 [office]
Important dates:
May 16, 1989 E-mail and papers must be received
July 3, 1989 Acceptance notifications mailed
July 25, 1989 Revised papers and single-page descriptions must be received
August 20, 1989 Date of Workshop
------------------------------
End of NL-KR Digest
*******************
