E1AR0002@SMUVM1.BITNET (Leff, Southern Methodist University) (09/22/87)
%T Expressiveness and tractability in knowledge representation and reasoning %A Hector J. Levesque %A Ronald J. Brachmand %J Computational Intelligence %V 3 %N 2 %D MAY 1987 %X A fundamental computational limit on automated reasoning and its effect on knowledge representation is examined. Basically, the problem is that it can be more difficult to reason correctly with one representational language than with another and, moreover, that this difficulty increases dramatically as the expressive power of the language increases. This leads to a tradeoff between the expressiveness of a representational language and its computational tractability. Here we show that this tradeoff can be seen to underlie the differences among a number of existing representational formalisms, in addition to motivating many of the current research issues in knowledge representation. %T Go\*:del, Lucas, and mechanical models of the mind %A Robert F. Hadley %J Computational Intelligence %V 3 %N 2 %D MAY 1987 %X In \fIMinds, Machines, and Go\*:del\fP, J.R. Lucas offers an argument, based on Go\*:del's incompleteness theorems, that the mind cannot be modeled by a machine. This argument has generated a variety of alleged refutations, some of which are incompatible with others. It is argued here that the incompatibility of these `refutations' points to a central paradox which has not yet been resolved. A solution to this paradox is presented, and a related paradox, concerning the existence of consistent models for inconsistent humans, is described and solved. An argument is presented to demonstrate that although humans commonly produce inconsistent output, they can, in an important sense, be modeled by \fIconsistent\fP formal systems, if their behavior is deterministic. It is also shown that Go\*:del's results present no obstacle to humans' proving the consistency of their own formal models. %T Domain circumscription: A re-evaluation %A David W. Etherington %A Robert Mercer %J Computational Intelligence %V 3 %N 2 %D MAY 1987 %X Some time ago, McCarthy developed the domain circumscription formalism for closed-world reasoning. Recently, attention has been directed towards other circumscriptive formalisms. The best-known of these, predicate and formula circumscription, cannot be used to produce domain-closure axioms; nor does it appear likely that the other forms can. Since these axioms are important in deductive database theory (and elsewhere), and since domain circumscription often can conjecture these axioms, there is reason to resurrect domain circumscription. .sp Davis presents an intuitively appealing semantics for domain circumscription, based on minimal models. However, under certain conditions McCarthy's syntactic realization of domain circumscription can induce inconsistencies in consistent theories with minimal models. We present a simple, easily motivated change that corrects this problem but retains the appealing semantics outlined by Davis. We also explore some of the repercussions of this semantics, including soundness and limited completeness results. %T Defeat among arguments: A system of defeasible inference %A R.P. Loui %J Computational Intelligence %V 3 %N 2 %D MAY 1987 %X This paper presents a system of non-monotonic reasoning with defeasible rules. The advantage of such a system is that many multiple extension problems can be solved without additional explicit knowledge; ordering competing extensions can be done in a natural and defensible way, via syntactic considerations. The objectives closely resemble Poole's objectives, but the logic is different from Poole's. The most important difference is that this system allows the kind of chaining that many other non-monotonic systems allow. Also, the form in which the inference system is presented is quite unusual. It mimics an established system of inductive logic, and it treats defeat in the way of the epistemologist-philosophers. The contributions are both of content and of form: the kinds of defeat that are considered, and the way in which defeat is treated in the rules of inference. %T A hybrid, decidable, logic-based knowledge representation system %A Peter F. Patel-Schneider %J Computational Intelligence %V 3 %N 2 %D MAY 1987 %X The major problem with using standard first-order logic as a basis for knowledge representation systems is its undecidability. A variant of first-order tautological entailment, a simple version of relevance logic, has been developed that has decidable inference and thus overcomes this problem. However, this logic is too weak for knowledge representation and must be strengthened. One way to strengthen the logic is create a hybrid logic by adding a terminological reasoner. This must be done with care to retain the decidability of the logic as well as its reasonable semantics. The result, a stronger decidable logic, is used in the design of a hybrid, decidable, logic-based knowledge representation system. %T Patterns of interaction in rule-based expert system programming %A Stan Raatz %A George Drastal %J Computational Intelligence %V 3 %N 2 %D MAY 1987 %X We study the effect of adding a rule to a rule-based heuristic classification expert system, in particular, a rule which causes an unforeseen interaction with rules already in the rule set. We show that it is possible for such an interaction to occur between \fIsets\fP of rules, even when no interaction is present between any \fIpair\fP of rules contained in these sets. A method is presented that identifies interactions between sets of rules, and an analysis is given which relates these interactions to rule-based programming practices which help to maintain the integrity of the knowledge base. We argue that the method is practical given some reasonable assumptions on the knowledge base. %A Charles Babcock %T IBM Expert Program Afforded Product Status %J ComputerWorld %P 118 %D JUL 13, 1987 %K AT02 %X IBM upgraded its expert system product from "introductory program to "full-fledged" product. It also has the capability of accessing it's "relational data base management systems." The complete system sells for $42,500. %T CAE software %J Electronic News %D July 6, 1987 %P 30 %V 33 %N 1662 %K AT02 AA05 %X Trimeter technologies has introduced a "knowledge-base" system to optimize ASIC designs costing $30,000. %T Kurzweil's Entry in Lowe-End Scanners %J Electronics %D JUN 11, 1987 %P 105 %V 60 %N 12 %K AT02 AI06 %X This $10,000 unit can read 60 characters/second, handle multiple type styles on the same page. It has a learning mechanism and a 10 to 40 thousand word lexicon. %T Fingerprint Reader Restricts Access to Terminals and PC's %J Electronics %D JUN 11, 1987 %P 104 %V 60 %N 12 %K H01 AT02 AI06 %X ThumbScan costs %995.00. .br ThumbScan Inc. Two Mid America Plaza, Suite 800, Oake Brook Terrace, Ill. 60181, 312-954-2336 %T This System Integrates DSP and Image Processor %J Electronics %D JUN 11, 1987 %P 106 %V 60 %N 12 %K AT02 AI06 %X Dataube integrates a Digital Signal Processor based on Analog Devices ADSP 2100 chip. It also contains video signal to bit conversion software. It also contains various hardware assists such as convolution. %T A new Way to Speed Up Artificial-Vision Systems %J Electronics %D JUN 11, 1987 %P 89-90 %V 60 %N 12 %K AI06 AT02 %X International Robotmotion's new image processing box contains multiple boards, each optimized for specific vision processes such as correlation, pixel statistical processor. It also has two on board array processors. The system costs $150,000. %A G. J. Holzmann %T Automated Protocol Validation in Argos: Assertion Proving and Scatter Searching %J IEEE Transactions on Software Engineering %D JUN 1987 %V SE-13 %N 6 %P 683-696 %K AA08 %A H. Gallaire %A J. Minker %A J. Nicolas %T Logic and Databases: A Response %J SIGPLAN Notices %V 22 %N 6 %D JUN 1987 %P 20-24 %A R. A. Sosnowski %T Prolog Dialects: A Deja Vu of BASICS %J SIGPLAN Notices %V 22 %N 6 %D JUN 1987 %P 39-48 %K T02 %X divides two Prolog styles into Edinburgh Prolog and micro-Prolog. Shows examples for various differences between these Prologs. Also discusses Turbo Prolog which he claims is still another dialect of Prolog. %A A. Cheese %T Multi-Moded Relations in Parlog %J SIGPLAN Notices %V 22 %N 6 %D JUN 1987 %P 49-51 %K T02 H03 %T New Entries Mark AI Shift From Lab to Market %J Electronic News %D JUL 20, 1987 %V 33 %N 1664 %K AT02 Data General Neuron Data DEC Digital Equipment Corporation T03 T01 Gensym O03 %X Discusses the following new products that appeared at the AAAI show: .br DEC new version of VAX VMS Lisp .br Package to support interchange of applications between personal computers and Data General MV models .br 386 board for Symbolics machines %T Ansa Brings Out Multi-User Version of Paradox DBMS %J Electronic News %V 33 %N 1661 %D JUN 29, 1987 %P 19 %K AA09 H01 AT02 %X Paradox 2.0 runs on various networks and supports complete record locking. %J ComputerWorld %D JUN 29, 1987 %V 21 %N 26 %P 20 %K AT12 AI01 %X response to Henry Eric Firdman's letter on how not to build an expert system. This letter states that the Dipmeter Advisor cost two million to build including costs associated with transferring to field use. %A Louis Fried %T The Dangers of Dabbling in Expert Systems %J ComputerWorld %D JUN 29, 1987 %V 21 %N 26 %K AI01 %X The SRI survey indicates the cost of application development for expert systems is $700 dollars per rule and this excludes hardware, software tools and the time of domain experts. The average cost is $260,000 per application. Goes on to discuss the importance of feasibility studies prior to building an expert system. Also discusses various characteristics of appropriate projects. %A David A. Ludlum %T Consortium set to Create Expert System Shell %J ComputerWorld %D JUN 29, 1987 %V 21 %N 26 %P 73 %K T02 AI02 Intellect AA09 AA06 AI01 AT16 %X Liberty Mutual Insurance Co., Southern California Edison, Transamerican Insurance and one other will be joining together to develop an expert system that can interface with mainframe DBMS and CICS. The system will use Intellect to formulate English queries to either DBMS or the rules themselves. %J ComputerWorld %V 21 %N 24 %P 25 %D JUN 15, 1987 %K AT04 %X France, Italy, UK and West Germany spent eighty million on AI software in 1986 and are expected to spend 825 million by 1991. %J InfoWorld %D JULY 6, 1987 %V 9 %N 27 %P 18 %K AT02 H01 T01 %X Star Saphire converts LISP to C. The resulting C code can be translated, compiled, optimized or link with other USER applications. It costs $495.00. %A Robert X. Cringeley %T Do They Really Want to Be as Smart as AT&T %J InfoWorld %D JULY 6, 1987 %V 9 %N 27 %P 78 %K AA08 %X AT&T plans to add AI tool to its network manager sometime in 1988. %T Canada Firm Set to Buy Lisp for $3.2M %J Electronic News %V 33 %N 1657 %D JUN 1, 1987 %P 12+ %K AT16 %X GigaMos Holdings has bought the assets of Lisp Machine Inc. (which earlier went into bankruptcy). GigaMos is affiliated with Lisp Canada. %A Karen Fitzgerald %A Paul Wallich %T Next Generation Race Bogs Down %J IEEE Spectrum %V 24 %N 6 %P 34-39 %K GA01 GA02 GA03 %X An NSF Team to assess the Japanese Fifth Generation came bakc with mixed conclusions: Japan is already ahead in certain area while others said that the Fifth Generation Project is a national embarassment. Marc Snir of the Hebrew University in Jerusalem said that there were many there because they were sent by their companies and that there was little questioning of efforts. An assessment of the ICOT Personel Sequential Inference Machine said that the system is inferior to US Lisp Workstations but has enormous physical memory (80 megabytes). .sp The article has a table of the various projects, Fifth Generation Computers, Alvey, Esprit, MCC and Strategic Computing, their goals, accomplishments and funding. %A Douglas Barney %T Microsoft in Link Pact %J ComputerWorld %V 21 %N 22 %D JUN 1, 1987 %P 8 %K AI02 H01 AT16 %X Microsoft licensed a natural language interface but no products are planned immediately. %T New Products %J ComputerWorld %V 21 %N 22 %D JUN 1, 1987 %P 34 %K Pyramid T01 AT02 %X Pyramid Announced a Pyrlisp system for $6000.00 %T New Products %J ComputerWorld %V 21 %N 22 %D JUN 1, 1987 %P 48 %K AI06 AT02 %X IBASE system reads documents and includes automatic form processing. %T Nickels and Dimes %J ComputerWorld %V 21 %N 22 %D JUN 1, 1987 %P 108 %K H02 %X Symbolics third quarter revenues ending March 29 was 24.6 million. %A E. Sacks %T Qualitative Sketching of Parameterized Functions %B Knowledge Based Expert Systems for Engineering: Classification, Education and Control %E D. Sriram %E R. A. Edey %I Computational Mechanics Institute %C Boston, USA %D 1987 %P 1-13 %K AA11 AA12 AA13 AI01 AI14 AA01 AA05 %X ISBN 0-931215-81-1 (Boston) ISBN 0-905451-92-9 (Southampton) %X This system uses Macsyma and "Bounder," a system that computes bounds from inequalities to do qualitative sketching of univariate expressions. It finds interesting points such as inflections, maxima, minima and discontinuities. QS has been tested on the following sets of equations in the literature: the four important probability distributions: uniform, exponential, gamma and normal, 17 out of 18 cases from the examples and exercises in Keeney and Raiffa's text on utility theory. Additional work will be done to deal with phase diagrams. %A J. Geller %A M. R. Taie %A S. C. Shapiro %A S. N. Srihari %T Device Representation and Graphics Interfaces of VMES %B Knowledge Based Expert Systems for Engineering: Classification, Education and Control %E D. Sriram %E R. A. Edey %I Computational Mechanics Institute %C Boston, USA %D 1987 %P 15-28 %K AI01 AA21 AI16 AA04 AI02 %X ISBN 0-931215-81-1 (Boston) ISBN 0-905451-92-9 (Southampton) %X Discusses representations of electronic systems to be maintained. Issues are the representation between the logical structure of the device and the physical entity of what is on a circuit board or other module to be replaced, graphical representation and natural language interface. %A D. J. Cooper %T An Expert Systems Approach to Process Identification and Adaptive Control %B Knowledge Based Expert Systems for Engineering: Classification, Education and Control %E D. Sriram %E R. A. Edey %I Computational Mechanics Institute %C Boston, USA %D 1987 %P 29-41 %K H01 AA20 AI01 T01 %X ISBN 0-931215-81-1 (Boston) ISBN 0-905451-92-9 (Southampton) %X Discusses a rule based system for adaptive control. The implementation has not been completed. They intend to write a Lisp-Fortran based system. %A R. H. Allen %A S. Haran %A V. Sharma %A J. Sorab %T Engineering and Artificial Intelligence Applications for the Evaluation and Management of Shoulder Dystocia %B Knowledge Based Expert Systems for Engineering: Classification, Education and Control %E D. Sriram %E R. A. Edey %I Computational Mechanics Institute %C Boston, USA %D 1987 %P 44-53 %K AA01 obstetrics delivery %X ISBN 0-931215-81-1 (Boston) ISBN 0-905451-92-9 (Southampton) %X This is a system to assist in the delivery of infants where the shoulder jams against the pelvic bone. A physical model, as well as a finite element model have been developed to assist in learning about the physical forces involved. In addition, a tactile sensing system to be worn underneath the physician's surgical glove was built for the purpose of measuring important parameters. The expert systems built as part of this effort include one to predict the possibility of this condition and another to manage it when it occurs. It is hoped that the other work on finite elements, tactile sensing and other modeling will be integrated within the system. This is an example of an expert system being build contemporaneously with the research to acquire the data on which it will be based. %T TAKING ISSUE: %T A critique of pure reason %J Computational Intelligence %V 3 %N 3 %D AUG 1987 %A Drew McDermott %X with peer commentary edited by Hector Levesque .sp The relevance of logic to AI has been hotly debated from the very beginnings of the field. Just when the issue seemed to be finally cooling down, Drew McDermott, a noted researcher and hitherto loyal advocate of logic, wrote a paper explaining why, after a decade of research, he has changed his mind about the use of logic. The special section of /Computational Intelligence/ will examine this issue in detail. After a short introduction, the section will contain McDermott's paper, together with commentaries on it by a number of prominent AI researchers: James Allen and Henry Kautz, Danny Bobrow and Mark Stefik, Ken Bowen, Ron Brachman, Eugene Charniak, Johan de Kleer, Jon Doyle, Ken Forbus, Pat Hayes, Carl Hewitt, Robert Kowalski, Robert Moore, Geoff Hinton, Jerry Hobbs, David Israel, John McCarthy and Vladimir Lifschitz, Nils Nilsson, Sandy Pentland, David Poole, Ray Reiter, Stan Rosenschein, Len Schubert, Brian Smith, Mark Stickel and Mabry Tyson, Richard Waldinger, Terry Winograd, and Bill Woods. Finally, McDermott replies to his critics. %T Equivalent logic programs and symmetric homogeneous forms of logic programs with equality %A Kwok-Hung Chan %J Computational Intelligence %V 3 %N 3 %D AUG 1987 %X This article introduces the notion of CAS-equivalent logic programs: logic programs with identical Correct Answer Substitution. It is shown that the notions CAS-equivalence, refutational equivalence, and logical equivalence do not coincide in the case of definite clause logic programs. Least-model criteria for refutational and CAS-equivalence are suggested and their correctness is proved. The least-model approach is illustrated by two proofs of CAS-equivalence. It is shown that the symmetric extension of a logic program subsumes the symmetry axiom, and the symmetric homogeneous form of a logic program with equality subsumes the symmetry, transitivity, and predicate substitutivity axioms of equality. These results contribute towards the goal of building equality into Standard Prolog without introducing additional inference rules. %T Pragmatic modeling: Toward a robust natural language interface %A M. Sandra Carberry %J Computational Intelligence %V 3 %N 3 %D AUG 1987 %X One of the most important ways in which an information-provider assimilates a n information-seeking dialogue is by inferring the underlying task-related plan motivating the information-seeker's queries. This paper presents a strategy for hypothesizing and tracking the changing task-level goals of an information-seeker, and building a model of his task-related plan as the dialogue progresses. .sp Naturally occurring utterances are often imperfect. The information-provider often appears to use inferred knowledge about the information-seeker's underlying task-related plan to remedy any of his faulty utterances and enable the dialogue to continue without interruption. This paper presents a strategy for understanding one kind of defective utterance. Our approach relies on the information-seeker's inferred task-related plan as the primary mechanism for suggesting how an utterance should be understood, thereby considering only interpretations that are relevant to what the information-seeker is trying to accomplish.