Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Posting-Version: version B 2.10.3 4.3bsd-beta 6/6/85; site ucbvax.BERKELEY.EDU Path: utzoo!decvax!decwrl!ucbvax!ailist From: AIList-REQUEST@SRI-AI.ARPA (AIList Moderator Kenneth Laws) Newsgroups: mod.ai Subject: AIList Digest V3 #180 Message-ID: <8511280054.AA09080@ucbvax.berkeley.edu> Date: Wed, 27-Nov-85 18:44:00 EST Article-I.D.: ucbvax.8511280054.AA09080 Posted: Wed Nov 27 18:44:00 1985 Date-Received: Thu, 28-Nov-85 19:25:34 EST Sender: usenet@ucbvax.BERKELEY.EDU Reply-To: AIList@SRI-AI Organization: The ARPA Internet Lines: 448 Approved: ailist@sri-ai.arpa AIList Digest Thursday, 28 Nov 1985 Volume 3 : Issue 180 Today's Topics: Queries - AI Workstations & Expert System on COBOL, Science Fiction - Machines That Talk, Humor - Doonesbury on Artificial Intelligence, Expert Systems - Liability & ES Strategies Newsletter, Literature - DNA Analysis & Recent Tech Reports, Seminar - The Riddle of STRIPS (SU) ---------------------------------------------------------------------- Date: 26 NOV 85 12:08-N From: MANUEL%CGEUGE52.BITNET@WISCVM.ARPA Subject: AI Workstations We are looking for an AI workstation for research purposes and would appreciate any comments that you have concerning your favorite (or not so favorite) workstation. In particular, Sun vs. Symbolics arguments would be appreciated. Here are some of the workstations that we are considering: Apollo DOMAIN family Symbolics 3600 DEC VAXstation II Tektronix TEK 4406 LMI Lambda family TI Explorer Sun 2 & 3 Xerox 1100 series Please send your comments directly to me. If you would like the contest results posted then please let me know. Thanks very much for your consideration. ... James Stewart Department of Physics University of Geneva MANUEL @ CGEUGE52 (on BITNET) ------------------------------ Date: Mon 25 Nov 85 13:10:38-EST From: Marc Vilain Subject: Expert system defeats halting problem? I'd be curious to know just which kinds of infinite loops are detected by IBM's COBOL restructuring program. No doubt that some cases can be detected by rudimentary kinds of template matching, but in general what we're talking about is the halting problem. Anybody out there have more information on this? marc. ------------------------------ Date: Sat, 23 Nov 85 10:59:02 GMT From: gcj%qmc-ori.uucp@ucl-cs.arpa Subject: Re: Fictional accounts of machines that talk (Vol 3 # 172) The natural language capabilities of a machine are amply illustrated in Robert Heinlein's ``The Moon is a Harsh Mistress'', (cf AIList Vol 3 # 39), written in the sixties. Here the machine, called Mike, talks to the programmer who realizes that Mike is now large enough to be self-aware. Not only does Mike hold quite normal conversations, he eventually creates a fictional character ``inside his head'', called Adam Selene. He then gives this character life by building a video and voice image. Many people would like to meet Mr. Selene in person! Also worth a mention in this context is ``Do Androids Dream of Electric Sheep?'', by Philip K. Dick. These are not quite the ancient Greeks, or particularly early. Gordon Joly gcj%qmc-ori@ucl-cs.arpa ------------------------------ Date: Sun 24 Nov 85 17:14:02-CST From: Werner Uhrig Subject: Doonesbury on Artificial Intelligence [ Doonesbury - Sunday Nov 24, 1985 ] [ Mike and Bernie, in front of a desk with a MACish computer facing them ] B: You see Mike, most of the computer breakthroughs we're making today have to do with Artificial Intelligence. The computer recognizes patterns in the user's behavior that enable it to make decisions on its own. Of course, the machine is still only as ethical as its owner. Let me show you a hypothetical model. Say I called up my company spread- sheet and transferred $100,000 to my personal account, okay? Watch. >BING!< EMBEZZLEMENT TRANSACTION COMPLETED. M: Good Lord !!! B: Mindless compliance, right? ... RESERVATIONS FOR RIO CONFIRMED. B: ... but it anticipated my every need ! ------------------------------ Date: Sun 24 Nov 85 12:30:41-PST From: PEREIRA@SRI-AI.ARPA Subject: Re: Expert Systems and Liability Given current litigation practices, I would think the answer to your questions is rather simple: you sue everyone in sight (doctor, program seller, program developer, hospital, city, state, Santa Klaus,...). With the ``joint and several'' legal doctrine currently used in liability cases, at least one of them will end up paying... (unless all go bankrupt, of course...) -- Fernando ------------------------------ Date: Wed 26 Nov 85 12:30:41-PST From: Laws@SRI-AI.ARPA Subject: Expert Systems Strategies Newsletter I have received a flyer for another AI newletter, Expert Systems Strategies. Paul Harmon is one of the editors, and his AI in Business book (with David King) is offered free to charter subscribers. The newsletter covers the usual mix: market trends, user profiles, vendor critiques, systems application analyses, performance engineering, hardware reviews, languages and tools, survey reports, knowledge engineering, events calendar, show updates, literature reviews, major contract announcements, and new business ventures -- all focussed on expert systems rather than AI in general. Current price is $207 ($40 off) for 12 monthly issues of 16 pages each. Cahners Newsletter Center, P.O. Box 59, New Town Branch, Boston, MA 02258; (617) 964-3030; Cable/CAHNERS BOSTON; Telex 94-0573 CPC BSN. ------------------------------ Date: Mon 25 Nov 85 01:23:34-CST From: Werner Uhrig Subject: interesting article in CACM of 85/11 1164 DISCOVERING THE SECRETS OF DNA. Peter Friedland and Laurence H Kedes Symbolic Pattern Recognition and the AI methodologies model building and theory formation will be critical to the next stage of discovery in regulatory molecular genetics. ------------------------------ Date: 24 Nov 1985 19:55-CST From: leff%smu.csnet@CSNET-RELAY.ARPA Subject: Recent Tech Reports ADDRESSES TO REQUEST TECH REPORTS LISTED: Department of Computer Science, Oregon State University, Corvallis, OR 97331 Department of Computer Science, Univeristy of Illinois at Urbana-Champign 1304 West Springfield Avenue, Urbana, Illinois 61801 OR erna@uiuc OR Engineering Docurments Center, University of Illinois at Urbana-Champaign, 1308 West Green Street, Urbana, Illinois 61801 Washington State University, Computer Science Deparmtent, Pullman, Washington 99164-1210 Computing Research Laboratory, University of Michigan, Room 1079, East Engineering Building, Ann Arbor, Michigan, 48109, USA UCLA Computer Science Department, 3732 Boelter Hall, Los Angeles, CA 90024 %A Asya Campbell %T Comparison of Dynamically and Statically Scoped Lisp %R CSD-850024 %I UCLA Computer Science Department %X $2.75 %A Rina Dechhter %T Studies in the Use and Generation of Heuristics %R CSD-850033 %I UCLA Computer Science Department %X $12.50 (Ph. D. Thesis) %K A* %A Judea Pearl %T Bayesian Networks: A Model of Self-Activiated Memory for Evidential Reasoning %R CSD-850021 %I UCLA Computer Science Department %X $0.50 %A Judea Pearl %T A Constraint-Propagation Approach to Probabilistic Reasoning %R CSD-850020 %I UCLA Computer Science Department %X $0.75 %A Judea Pearl %T Fusion, Propagation, and Structuring in Bayesian Networks %R CSD-850022 %I UCLA Computer Science Department %X $4.25 %A Judea Pearl %T Bayes Decision Methods %R CSD-850023 %I UCLA Computer Science Department %X $2.00 %A Judea Pearl %A Michael Taria %T Structuring Causal Trees %R CSD-850029 %I UCLA Computer Science Department %X $1.75 %A Judea Pearl %T How to Do With Probabilities What People Say You Can't %R CSD-850031 %X $1.75 %I UCLA Computer Science Department %X $1.75 %A Judea Pearl %T On Evidential Reasoning in a Hierarchy of Hypotheses %R CSD-850032 %X $0.75 %I UCLA Computer Science Department %A Michael A. Langston %A Chul E. Kim %T Movement Coordination for Single-Track Robot Systems %R CSD-84-125 %X $2.50 (discusses dealing with multiple robots) %I Washington State University Computer Science Department %A Jerzy Tiuryn %T An Introduction to First-Order Programming Logics %I Washington State University Computer Science Department %R CSD-84-126 %X $4.20 %A Keshav Sharma %T Syntactic Aspects of the Non-Deterministic Lambda Calculus %R CSD-84-127 %I Washington State University Computer Science Department %X $7.20 %A Jerzy Tiuryn %T A Simplified Proof of DDL < DL %R CS-85-130 %I Washington State University Computer Science Department %X 1.50 (compares Deterministic Dynamic Logics and Dynamic Logics of regular programs) %A David Matthew Dahlbacka %T An ATN-Based Restricted Natural Language Front End for A Data-Flow Design Aid %R Department of Computer Science File No. 944 %I University of Illinois at Urbana-Champaign %D JUL 1985 %K software engineering %A Tomoyasu-Taguti Nakagawa %A Hung-Chi Lai %T Reference Manual of Fortran Program ILLOD-(NOR-B) for Optimal NOR Networks %R Department of Computer Science Report No. 1129 %I University of Illinois at Urbana-Champaign %D JUL 1985 %K branch and bound %A Simon M. Kaplan %T Verification of Recursive Programs: A Temporal Proof Approach %R Department of Computer Science Report NO. 1207 %I University of Illinois at Urbana-Champaign %D SEP 1985 %A Larry Rendell %T Induction, Of and By Probability %R Department of Computer Science Report NO. 1209 %I University of Illinois %D JUL 1985 %X This paper examines some methods and ideas underlying the author's successful probabilistic learning systems (PLS). These systems have proven uniquely effective and efficent for generalization learning (induction) in heuristic search. Aspects of PLS include use of probabilities to guide both task performance and learning, incremental revision and normalization of probabilities, and localization and correction of their errors. Construction of new terms (features) for heuristic functions may be feasible. %A Larry Rendell %T Genetic Plans and the Probabilistic Learning System: Synthesis and Results %R Department of Computer Science Report NO. 1217 %I University of Illinois at Urbana-Champaign %D JUL 1985 %X describes PLS2 which clusters data into economical cells in augmented feature space, and a genetic level which selects successful regions by a genetic algorithm. %A Nachum Dershowitz %T Termination of Rewriting %R Department of Computer Science Report NO. 1220 %I University of Illinois at Urbana-Champaign %D AUG 1985 %A Jean-Luc Remy %T Review of Several Closure Properties in Universal Algebra and First Order Logic %R Department of Computer Science Report NO. 1221 %I Univeristy of Illinois at Urbana-Champaign %D JUL 1985 %A John Shilling %T Initial Report on ISADORE: A Reference Librarian Generator %R Department of Computer Science Report NO. 1225 %I University of Illinois at Urbana-Champaign %D SEP 1985 %A S. R. Ray %A W. D. Lee %A C. D. Morgan %A W. Airth-Kindree %T Computer Sleep Stage Scoring - An Expert System Approach %R Department of Computer Science Report No. 1228 %I University of Illinois at Urbana-Champaign %D September 1985 %A T. G. Lewwis %T An Operator Calculus for Computer Programs %R CSTR-78-1-5 %I Oregon State University, Department of Computer Science %D 1978 %A W. S. Bregar %A A. M. Farley %T Interactive Problem Solving in Elementary Algebra %R CSTR-78-3-1 %I Oregon State University, Department of Computer Science %D 1978 %A P. Cull %A W. Frank %T Flaws of Form %D 1978 %R CSTR-78-20-3 %I Oregon State University, Department of Computer Science %X G. Spencer Brown's book \fBLaws of Form\fR has been enjoying a vogue among social and biological scientists. Proponents claim that the book introduces a new logic ideally suited to their fields of study, and that the new logic solves the problems of self-reference. These claims are false. We show that Brown's system is Boolean algebra in an obscure notation, and that his "solutions" to the problems of self-reference are based on a misunderstanding of Russell's paradox. %A B. Levin %T The Automated Inference of Tree Systems %D 1979 %R CSTR-79-20-6 %I Oregon State University, Department of Computer Science %A B. Blanchard %A W. S. Bregar %T A Production Environment %D 1982 %R CSTR-82-3-1 %I Oregon State University, Department of Computer Science %A D. Kogan %A M. J. Freiling %T SIDUR: A Structuring Formalism for Knowledge Information Processing Systems %D 1984 %R CSTR-84-40-2 %I Oregon State University, Department of Computer Science %A S. Rehfuss %A M. J. Freiling %A J. Alexander %T Particularity in Engineering Data %D 1984 %R CSTR-84-40-3 %I Oregon State University, Department of Computer Science %X discusses inferential data in three engineering expert system databases %A Soveig A. Viste %A Chul E. Kim %T The Recognition of Digital Cylindrical Surfaces and Digital Moebius Strips %R CS-85-133 %I Washington State University Deparment of Computer Science %X $3.50 %A Y. Gurevich %A S. Shelah %T Fixed-Point Extension of First-Order Logic %R CRL-TR-5-85 %I Computing Research Laboratory, University of Michigan %D May 1985 ------------------------------ Date: 22 Nov 85 1432 PST From: Vladimir Lifschitz Subject: Seminar - The Riddle of STRIPS (SU) The Riddle of STRIPS and Its Solution Vladimir Lifschitz Non-Monotonic Reasoning Seminar Wednesday, November 27, 2pm MJH 252 STRIPS (STanford Research Institute Planning System) operates with world models represented by sets of formulas of first-order logic. A STRIPS system describes the effect of an action by a rule which defines how the current world model should be changed when the action is performed. The big mystery about STRIPS is why it does not produce incorrect results. Presumably, this happens because the rules correctly describe properties of the corresponding actions. But what do we mean by the "correctness" of a STRIPS rule? Straightforward attempts to define the semantics of STRIPS rules turn out to be unsatisfactory; we will examine a classical STRIPS system and show that, from some points of view, its rules are incorrect. The purpose of that exercise is not to criticize the system, but rather to show that defining the semantics of STRIPS is a tricky business. In the last part of the talk, a solution to the problem will be proposed. We will see that, under some conditions, STRIPS rules can be viewed as perfectly legitimate tools for formalizing knowledge about the effects of actions. ------------------------------ End of AIList Digest ********************