Showing papers on "Representation (systemics) published in 1989"

A model for reasoning about persistence and causation

Abstract: In this paper, we describe a model of causal reasoning that accounts for knowledge concerning cause-and-effect relationships and knowledge concerning the tendency for propositions to persist or not as a function of time passing. Our model has a natural encoding in the form of a network representation for probabilistic models. We explore the computational properties of our model by considering recent advances in computing the consequences of models encoded in this network representation.

Similarity, typicality, and categorization

Abstract: Here is a simple and appealing idea about the way people decide whether an object belongs to a category: The object is a member of the category if it is sufficiently similar to known category members. To put this in more cognitive terms, if you want to know whether an object is a category member, start with a representation of the object and a representation of the potential category. Then determine the similarity of the object representation to the category representation. If this similarity value is high enough, then the object belongs to the category; otherwise, it does not. For example, suppose you come across a white three-dimensional object with an elliptical profile; or suppose you read or hear a description like the one I just gave you. You can calculate a measure of the similarity between your mental representation of this object and your prior representation of categories it might fit into. Depending on the outcome of this calculation, you might decide that similarity warrants calling the object an egg, perhaps, or a turnip or a Christmas ornament. This simple picture of categorizing seems intuitively right, especially in the context of pattern recognition. A specific egg – one you have never seen before – looks a lot like other eggs. It certainly looks more like eggs than it looks like members of most other categories. And so it is hard to escape the conclusion that something about this resemblance makes it an egg or, at least, makes us think it's one.

Meaning and mental representation

Abstract: In this provocative study, Robert Cummins takes on philosophers, both old and new, who pursue the question of mental representation as an abstraction, apart from the constraints of any particular theory or framework. Cummins asserts that mental representation is, in fact, a problem in the philosophy of science, a theoretical assumption that serves different explanatory roles within the different contexts of commonsense or "folk" psychology, orthodox computation, connectionism, or neuroscience. Cummins looks at existing and traditional accounts by Locke, Fodor, Dretske, Millikan, and others of the nature of mental representation and evaluates these accounts within the context of orthodox computational theories of cognition. He proposes that popular accounts of mental representation are inconsistent with the empirical assumptions of these models, which require an account of representation like that involved in mathematical modeling. In the final chapter he considers how mental representation might look in a connectionist context. A Bradford Book.

Database retrieval system having a natural language interface

Abstract: A database retrieval system having a natural language interface is provided. A database developer creates a knowledge base containing a structural description and semantic description of an application database from which data is to be retrieved. A database independent, canonical internal meaning representation of a natural language query is produced. An expert system accesses structural and semantic description information in the knowledge base and, in accordance with predefined rules, identifies database elements from said information that are necessary to satisfy the query represented by the internal meaning representation. A database query is generated among the database elements, enabling the retrieval and aggregation of data from the database to satisfy the natural language query. A debugging facility derives an external meaning representation from the internal meaning representation. The external meaning representation is database-independent, canonical, and easily understandable to the database developer. The external meaning representation enables the database developer to comprehend the internal meaning representation and verify that a natural language query is properly interpreted by the system to effect the accurate retrieval and aggregation of data from the database. The external meaning representation comprises entities and constraints relating to the entities, without reference to factual or linguistic relationships between entities that would prevent the external meaning representation from being easily understood.

Unification theory

Abstract: Most knowledge based systems in artificial intelligence (AI), with a commitment to asymbolic representation, support one basic operation: ''matching of descriptions''. This operation, called unification in work on deduction, is the ''addition-and-multiplication'' of AI-systems and is consequently often supported by special purpose hardware or by a fast instruction set on most AI-machines. Unification theory provides the formal framework for investigations into the properties of this operation. This article surveys what is presently known in unification theory and records its early history.

Adaptive information retrieval: using a connectionist representation to retrieve and learn about documents

Abstract: AIR represents a connectionist approach to the task of information retrieval. The system uses relevance feedback from its users to change its representation of authors, index terms and documents so that, over time, AIR improves at its task. The result is a representation of the consensual meaning of keywords and documents shared by some group of users. The central focus goal of this paper is to use our experience with AIR to highlight those characteristics of connectionist representations that make them particularly appropriate for IR applications. We argue that this associative representation is a natural generalization of traditional IR techniques, and that connectionist learning techniques are effective in this setting.

Functional representation of designs and redesign problem solving

Abstract: The in fo rmat ion processing task of redesign and i ts subtasks of diagnosis and repair are an­ alyzed. Various kinds of knowledge required for redesign problem solving are ident i f ied, and a scheme for representing them is described. In this scheme, the funct ions of the device and its s t ruc tura l components are represented ex­ p l ic i t ly , and causal and ant ic ipatory knowledge about its design is organized around these func­ t ions. Th is funct ional representation language also provides pr imi t ives for representing and ac­ cessing knowledge of domain principles such as Physics laws. The use of funct ional represen­ ta t ion of designs in redesign problem solving is i l lus t ra ted for the redesign of the reaction wheel assembly aboard the Hubble space telescope. 1 D e s i g n : P r o p o s a l , V e r i f i c a t i o n , and Redes ign The design problem can be abstract ly characterized as a constrained function-to-structure mapp ing . The de­ sign task takes as input the specifications of the desired funct ions of a device and the constraints on the design, and produces as ou tpu t a specif ication of a structure that realizes the desired funct ions and satisfies the con­ stra ints. One way to analyze a complex task such as design is to ident i fy the methods that can be appl ied to the task, the knowledge and contro l that these methods require, and the subtasks generated by them. Th is anal­ ysis produces a task structure [Chandrasekaran, 1989], i.e., a task-subtask decomposit ion of the prob lem, along w i t h a specif ication of the knowledge required for each of the subtasks. For a given task in this task st ructure, the choice of the method can depend on the knowledge available to the problem solver and the computa t iona l efficiency of f inding the solut ion by various methods ap­ plicable to the task. One method for solving design problems is propose, verify, and redesign [Chandrasekaran, 1988]. Th is method identifies and orders three subtasks, each of which in tu rn can be performed in different domains by different methods. For instance, case-based methods have recently become a subject of research for the propose subtask, and the verify subtask can be performed by a variety of methods, inc lud ing actual test ing of the device, analyt ic methods such as finite element analysis, and various s imulat ion techniques. One goal of this pa­ per is to perform an analysis of the task of redesign in terms of the subtasks into which it can be decomposed, and the methods applicable to them. The second goal of this paper is to explore the use of function-structure models for the redesign task, specifically, to investigate the u t i l i t y of the functional representation scheme Sembugamoor thy and Chandrasekaran, 1986 which models the relat ionship between the structure of a device, the behaviors tha t arise f rom i t , and the teleology of the device as a whole. Th is research builds on our earlier work Coel and Chandrasekaran, 1988] in which we pro­ posed the use of funct ional representation of designs in c r i t iqu ing a proposed design, i.e., in local izing the fai lure to deliver a funct ion to a part of the st ructure. "Th is research has been supported by the Defense Advanced Research Projects Agency, R A D C contract F30602-85-C-0010, the Nat ional Science Foundation, grant CBT-87-03745, and the McD onnell Douglas Corporat ion, research contract W S M D R L 2931. Computer facilit ies were enhanced by gifts f rom the Xerox Corporat ion. 1.1 R e d e s i g n : C o r r e c t i v e a n d C o m p e n s a t o r y Redesign is tr iggered as a task whenever the verify subtask shows that the proposed design falls short of the desired, either because some of the desired funct ions are not realized or because some of the behaviors are unde­ sirable. Once the proposed design has been modi f ied, the verify-redesign cycle is repeated if the design is get t ing closer to the desired one, or a different candidate design is sought f r om the propose subtask. In this paper, we are par t icu lar ly concerned w i t h redesign prob lem solv­ ing when the verify subtask finds an undesirable device behavior. The redesign of a bal l bearing assembly which generates excess heat due to large ro ta t iona l loads, where the generation of excess heat is an unintended and un­ desirable device behavior, is an example of this generic class of redesign problems. Solutions to this redesign problem can be corrective, or compensatory, or some combinat ion of the two. The redesigner may diagnose and repair the s t ruc tura l fau l t re­ sponsible for an undesirable behavior, or it may propose addi t ional structures that can compensate for the unde­ sirable behavior. If, for instance, isolat ing the s t ruc tura l faul t responsible for an undesirable behavior or fu l ly cor­ recting it is not feasible, or is computa t iona l l y too ex­ pensive, then the redesigner may devise a compensatory solut ion to the problem. In the bal l bearing example, the proposal for the use of a cooler to remove the excess heat generated is a compensatory redesign so lut ion. In this paper, we are especially interested in corrective redesign 1388 Knowledge Representation problem solving. The method of correcting an undesirable behavior fur­ ther decomposes the redesign task into two subtasks: diagnosis and repair. The diagnosis subtask takes the proposed structure and its undesirable behaviors as in­ put , and gives the st ructura l causes for the undesirable behaviors as the output . The repair subtask of redesign takes the desired funct ions, the proposed structure, the undesirable behaviors and their s t ructural causes as in­ put , and produces as ou tpu t a modif ied structure that realizes the desired functions w i thout the undesirable behavior. The diagnosis subtask can be performed by a variety of methods ranging f rom associative mapping of behavior to structure to techniques based on simu­ la t ion of behavior f rom structure. Below we present a method for diagnosis and repair that makes use of func­ t ional representations of designs in the form of stored structure-to-function maps. 2 Reac t i on W h e e l Assemb ly In order to make the present discussion more concrete, let us consider the specific problem of redesigning the reaction wheel assembly (RWA) aboard the Hubble space telescope, a slice of which is shown in Figure 1. The desired funct ion of RWA is to make the telescope point at a chosen area of the sky. The given structure of the RWA consists of a rapidly spinning rotor mounted on a shaft. The rotat ing shaft is connected to a stator at both ends via assemblies of ant i f r ic t ion ball bearings. The power that drives the rotor comes from a motor that is remotely control led f rom earth. The stator itself is mounted on the walls of the telescope bay. The constraint on the design of RWA is to keep its mass as small as possible. The funct ion ing of RWA is based on the law of conser­ vat ion of angular momentum. When the telescope is to be oriented in a specific d i rect ion, a signal f rom earth is sent to the motor that results in a change in the power supplied to the rotor. This causes a change in the angu­ lar velocity of the rotor and a corresponding change in its angular momentum. Due to the conservation of angu­ lar momentum, the angular momen tum of the telescope as a whole changes in the opposite direct ion. When the telescope nears its desired or ientat ion, a change in the angular momentum of the telescope in the opposite di­ rection is achieved in a similar manner, and the telescope angular velocity is reduced to zero. A common problem in the operat ion of RWA arises due to f r ic t ion in the bearing assemblies. The load on the bearings due to the rapid spin of the rotor causes de­ format ion of the bearing balls which results in increased f r ic t ional forces in the bearing assembly. This causes generation of heat in the bearing assembly. The resultins increase in temperature is detected by temperature sensors located near the bearing assemblies. Since the in­ crease in temperature depends on the load on the bear­ ings, a typical redesign solut ion to this problem is to increase the load capacity of the bearings by increasing the size of the balls. The increased temperature in the bearing assembly is an example of an unintended and undesirable behavior. The designer of HWA anticipated the potential for this undesirable behavior and included sensors in the design specifically to detect its presence. Note that because of the constraint of keeping the mass as small as possible, increasing the size of the bearing balls by an arbi t rar­ ily large amount is not an acceptable redesign solut ion. Also, since the effects of rotat ional loads on bearing as­ semblies are not known analyt ical ly, it is not possible to exactly compute the smallest size ball bearings that can support a given rotat ional load. In using the redesign of RWA as an i l lustrat ion of our analysis of redesign problem solving, we assume that the angular momentum of the telescope as a whole is in i t ia l l y zero, and that the angular momentum of the rotor is in the anticlockwise direct ion. We also assume that the command f rom earth is to increase the angular velocity of the rotor so that the telescope acquires an angular momentum in the opposite direct ion, and that the de­ sired thange in the magnitude of angular momentum is proport ional to the magnitude of the command signal. Whi le these assumptions reduce the size of the problem, thev do not entai l any loss of generality. 3 Func t iona l O rgan i za t i on of Des ign

Representation properties of networks: Kolmogorov's theorem is irrelevant

Abstract: Many neural networks can be regarded as attempting to approximate a multivariate function in terms of one-input one-output units. This note considers the problem of an exact representation of nonlinear mappings in terms of simpler functions of fewer variables. We review Kolmogorov's theorem on the representation of functions of several variables in terms of functions of one variable and show that it is irrelevant in the context of networks for learning.

Representation and Reasoning with Attributive Descriptions.

Abstract: This paper surveys terminological representation languages and feature-based unification grammars pointing out the similarities and differences between these two families of attributive description formalisms. Emphasis is given to the logical foundations of these formalisms.

The Iconography of Landscape: Essays on the Symbolic Representation, Design and Use of Past Environments

Abstract: Preface Introduction: iconography and landscape Stephen Daniels and Denis Cosgrove 1. The geography of Mother Nature Peter Fuller 2. The evocative symbolism of trees Douglas Davies 3. The political iconography of woodland in later Georgian England Stephen Daniels 4. Places and dwellings: Wordsworth, Clare and the anti-picturesque John Lacas 5. Art and agrarian change, 1710-1815 Hugh Prince 6. 'Fields of radiance': the scientific and industrial scenes of Joseph Wright David Fraser 7. The privation of history: Landseer, Victoria and the Highland myth Trevor P. Pringle 8. The iconography of nationhood in Canadian art Brian S. Osborne 9. Rhetoric of the western interior: modes of environmental description in American promotional literature of the nineteenth century G. Malcolm Lewis 10. Symbolism, 'ritualism' and the location of crowds in early nineteenth-century English towns Mark Harrison 11. Symbol of the Second Empire: cultural politics and the Paris Opera House Penelope Woolf 12. The sphinx in the north: egyptian influences on landscape, architecture and interior design in eighteenth- and nineteenth-century Scotland Eric Grant 13. The geometry of landscape: practical and speculative arts in sixteenth-century Venetian land territories Denis Cosgrove 14. Maps, knowledge, and power J. B. Harley Index.

A visual representation for knowledge structures

Abstract: Knowledge-based systems often represent their knowledge as a network of interrelated units. Such networks are commonly presented to the user as a diagram of nodes connected by lines. These diagrams have provided a powerful visual metaphor for knowledge representation. However, their complexity can easily become unmanageable as the knowledge base (KB) grows.This paper describes an alternate visual representation for navigating knowledge structures, based on a virtual museum metaphor. This representation uses nested boxes rather than linked nodes to represent relations. The intricate structure of the knowledge base is conveyed by a combination of position, size, color, and font cues, MUE (Museum Unit Editor) was implemented using this representation to provide a graphic front end for the Cyc knowledge base.

Abstract theorem proving

Abstract: Informally, abstraction can be described as the process of mapping a representation of a problem into a new representation. The aim of the paper is to propose a theory of abstraction. The generality of the framework is tested by formalizing and analyzing some work done in the past; its efficacy by giving a procedure which solves the false proof problem by avoiding the use of inconsistent abstract spaces.