nl-kr-request@cs.rpi.edu (NL-KR Moderator Chris Welty) (02/22/89)
NL-KR Digest (2/21/89 19:34:05) Volume 6 Number 3 Today's Topics: BBN AI Seminar -- Jim Schmolze BBN AI/Education seminar: Sayeki & Ueno SUNY Buffalo Cognitive Science Colloquium CSLI Calendar, Feb. 16 Submissions: nl-kr@cs.rpi.edu OR nl-kr@turing.cs.rpi.edu Requests, policy: nl-kr-request@cs.rpi.edu OR nl-kr-request@turing.cs.rpi.edu %% My priority will be to get the backlog of announcements out first, %% then the questions and discussions. This digest is composed of %% announcements that are still timely. I will post the announcements %% that are late (in case they are of interest anyway) in the next issue. ---------------------------------------------------------------------- Date: Tue 21 Feb 89 14:44:41-EST >From: Marc Vilain <MVILAIN@G.BBN.COM> Subject: BBN AI Seminar -- Jim Schmolze To: ai-folks@G.BBN.COM BBN Science Development Program AI Seminar Series Lecture GUARANTEEING SERIALIZABLE RESULTS IN PRODUCTION SYSTEMS THAT EXECUTE MANY RULES IN PARALLEL Jim Schmolze Department of Computer Science Tufts University (schmolze%cs.tufts.edu@RELAY.CS.NET) BBN Labs 10 Moulton Street 2nd floor large conference room 10:30 am, Tuesday February 28 To speed up production systems, researchers have studied how to execute many rules simultaneously. Unfortunately, such systems can yield results that are impossible for a serial system to produce, leading to erroneous behaviors. I will present a fast algorithm that prevents all non-serializable effects for multi-rule execution systems. My framework is taken from [IS85] and improves upon their solution. I also offer two additional strategies for increasing concurrency while retaining the serialization guarantee. The practical advantages of these strategies is considerable, as shown by my estimates for a large production system, the Manhattan Mapper [LC83]. [IS85] T. Ishida and S.J. Stolfo. "Towards the parallel execution of rules in production system programs." In Proceedings of the International Conference on Parallel Processing, 1985. [LC83] L. Lerner and J. Cheng. "The Manhattan Mapper expert production system". Technical report, Dept. of Computer Science, Columbia University, May 1983. ------------------------------ Date: Thu 16 Feb 89 11:23:36-EST >From: Marc Vilain <MVILAIN@G.BBN.COM> Subject: BBN AI/Education seminar: Sayeki & Ueno To: ai-folks@G.BBN.COM BBN Science Development Program AI Seminar Series Lecture MENTAL MODELS AS THEATER YUTAKA SAYEKI, NAOKI UENO, University of Tokyo National Institute of Educational Research BBN Labs 10 Moulton Street 2nd floor large conference room 10:30 am, Tuesday February 21 This talk will focus on the understanding physics problems, taking a model of the theater and combining it with kobito theory in which point of view has a critical role in understanding objects and environments. In Theater, as Peckham (1965) describes, we have an enormously rich variety of metaphors for new features for computer interfaces that can aid in understanding: Actors, Directors, Stage Conductors, Audience, Critics, Stage, Stage-Setting, Background, Foreground, Scenes, Play, , Casts, Casting, Script, Scripting, Rehearsal, Dramatist , Improvising, Ad-lib, Show, and so on. In kobito theory as elaborated by Sayeki, "point of view" and active participation in different modes of activities (such as "throwing in, " "acting out," and "feeling about") is considered to be crucial for exercising roles of actors, viewers (audience and directors) in order to get deep understanding. The following features are found in the notions elaborated by Peckham, Ueno, and Sayeki: (1) Every "view" must be a view from a particular vantage point in situ playing a particular Role, that can be shifted, moved, exchanged, or replaced. The important point here is is that we actively choose and try out taking a variety of vantage points, in order to delineate the critical "invariant structure" (cf. Gibson) of the scene. Shifting vantage points can yield "insight" into a solution as Peckham described, too. We need "tools" and "stages" for searching and trying-out possible vantage points. (2) We learn more from observing the continuous changes of scenes, or movements of objects along with our own movements and actions upon the objects, rather than fixed "representations" or "snapshots" of objects . (Here again, we take a Gibsonian view, rather than the "representationalist's view" of cognitive science.) (3) We learn and think by acting, participating, and changing in a broad domain of activity, rather than simply watching or manipulating objects in your hands or on your "desktop," without moving your original position. An important point here is that we occasionally change the domain of activity, such as working at the desk, travelling by car, train, and airplane, attending conferences, working at home at night, and so on. Current interface technology assumes an "armchair viewer" at the fixed position. (4) "Representation" is NOT a thought by itself; it is a medium of thoughts. "Representation" should be "social" from the beginning and be used "socially." It must be deeply rooted in cultural, "shared" knowledge, as well as triggered by the materialized "form" or appearance of the object to be represented. In the presentation, we will explain a number of misunderstandings of physics problems as either (1) miscasting of actors, or (2) mis-staging of the environment. Thus it would be possible to "cure" some of the "conceptual bugs" by re-casting or re-staging the situations. We shall illustrate these points by the use of 3D Logo. ------------------------------ Date: Fri, 17 Feb 89 10:16:46 EST >From: rapaport@cs.Buffalo.EDU (William J. Rapaport) To: nl-kr@turing.cs.rpi.edu, Subject: SUNY Buffalo Cognitive Science Colloquium UNIVERSITY AT BUFFALO STATE UNIVERSITY OF NEW YORK GRADUATE GROUP IN SEMIOTICS and GRADUATE RESEARCH INITIATIVE IN COGNITIVE AND LINGUISTIC SCIENCES PRESENT JAMES H. FETZER Deparment of Philosophy University of Minnesota at Duluth SEMIOTIC SYSTEMS AND ARTIFICIAL INTELLIGENCE Monday, February 27, 1989 4:00 P.M. 684 Baldy Hall, Amherst Campus There will be an evening discussion at 8:00 P.M. at Erwin Segal's house, 101 Carriage Circle, Williamsville. Contact Paul Garvin, Department of Linguistics, 636-2177, or Bill Rapaport, Department of Computer Science, 636-3193, for further information. ------------------------------ Date: Thu, 16 Feb 89 08:15:26 PST >From: hyde@csli.Stanford.EDU (Dawn Hyde) To: emma@csli.Stanford.EDU Subject: csli calendar, Feb. 16, 4:16 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 _____________________________________________________________________________ 16 February 1989 Stanford Vol. 4, No. 16 _____________________________________________________________________________ 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, 16 February 1989 12 noon TINLunch Cordura Hall Defeasible Reasoning and Nonmonotonic Conference Room (and worse) Inference Relations: A Little Philosophy; A Little AI; A Little Logic David Israel (israel@ai.sri.com) Abstract below 2:15 p.m. CSLI Seminar Cordura Hall Overview of the German Research Center for AI Conference Room Gerhard Barth German Research Center for AI Abstract below 3:30 p.m. Tea Ventura Hall ____________ THIS WEEK'S TINLUNCH Defeasible Reasoning and Nonmonotonic (and worse) Inference Relations: A Little Philosophy; A Little AI; A Little Logic David Israel 16 February Starting about a decade ago, researchers in Artificial Intelligence began to formalize certain oddly behaved, in particular nonmonotonic, inference patterns. Some of these, e.g., Reiter's logic for default reasoning, modeled features of actually existing computational systems; others, such as circumscription, were presented as capturing important features of actual, human common-sense reasoning. In both kinds of case, the phenomenon under consideration has something to do with what philosophers have called `defeasible reasoning.' The latter will be very briefly characterized. We shall then move on to recent attempts by logicians Dov Gabbay, on the one hand, and David Makinson, on the other, to give abstract axiomatic accounts of a variety of these nonmonotonic inference relations. These will be described and some important results mentioned. A few questions will be raised about what such inference relations have to do with defeasible reasoning. `Many' fewer answers will be suggested. ____________ THIS WEEK'S SEMINAR Overview of the German Research Center for AI Gerhard Barth German Research Center for AI 16 February The German Research Center for AI was founded just recently. In this talk, an overview of its structure, organization, and short-term research program will be presented. One of the research projects that has already been started is concerned with knowledge-based presentation of information. Some specific issues of this project will be discussed in the second half of the talk. ____________ LINGUISTICS DEPARTMENT COLLOQUIUM A Typology of Possible Morphophonological Change Bill J. Darden University of Chicago Friday, 17 February, 3:15 Cordura Hall Conference Room Morphonological rules, being phonologically unmotivated, can be seen as creating formal and semiotic problems. They have to be learned and they create unmotivated allomorphy. They also may serve as auxiliary signs themselves. Changes can generally be classified as those that solve the problems or those that enhance the sign value of the alternation. These changes indicate the close ties of morpho-phonology to morphology. Other changes indicate close ties with phonology. There are phonologically motivated adjustments in morphonological rules, and morphologically motivated adjustments in phonological rules (which may result in the elimination of the rules). Ultimately, however, the phonologically motivated changes can be interpreted as morphologically motivated, and the morphological influence on phonological rules can be limited to the morphological aspect (the input). _____________ SYMBOLIC SYSTEMS FORUM Semiotics David Wellbery German Studies Friday, 17 February, 3:15 Room 60:61G David Wellbery will explain why symbolic systems should consider the humanities as a part of its major. There has been much work on symbols, symbolism, meaning, interpretation, and much more in the humanities (principally semiotics, literary theory, and symbolic anthropology), which researchers and students in symbolic systems could use and might miss otherwise. On the other hand, as in every case of collaboration, these humanities disciplines also stand to gain great benefit from ideas within technical symbolic systems. In this vein, Professor Wellbery hopes to hold an informal discussion in which he will present some ideas about semiotics and attempt to justify collaboration. ___________ THE HISTORICAL LINGUISTICS INTEREST GROUP The Diachronic Typology of Possessives Dr. Vjacheslav V. Ivanov Professor of History of Culture, Moscow University Thursday, 23 February, 5:30 Ventura Hall Seminar Room Refreshments will be served at 5:00. ____________ SYMBOLIC SYSTEMS FORUM Turing's Oracle Solomon Feferman Department of Mathematics Friday, 24 February, 3:15 Room 60:62N ____________ LINGUISTICS DEPARTMENT COLLOQUIUM Cross-Linguistic Properties of Anaphoric Systems Peter Sells Department of Linguistics Friday, 24 February, 3:30 Cordura Hall Conference Room ____________ ------------------------------ End of NL-KR Digest *******************