A relational characterization of the space of models of complex systems for diagnostic problem solving

    Authors

    R. A. Morris; A. J. Gonzalez;D. J. Carreira

    Comments

    Authors: contact us about adding a copy of your work at STARS@ucf.edu

    Abbreviated Journal Title

    Int. J. Expert Syst.

    Keywords

    Computer Science, Artificial Intelligence

    Abstract

    The effective diagnosis of complex systems requires reasoning with multiple models of the system. Recent research in reasoning about complex systems has been directed towards the representation of, and reasoning with, multiple models. This paper addresses some representational issues that arise when a diagnostic problem solver is equipped with the means of reasoning with multiple models of a system. A theory is proposed to explain how such a problem solver is able to effectively select models for diagnosing the system. A necessary condition for effective reasoning with multiple models is the dynamic selection of models which are as simple as possible to fulfill the requirements of the task. Models are viewed here as corresponding to relations defined on a set of variables. This abstraction allows for a precise characterization of the space of possible models induced by a finite set of operations on a finite set of initial models, as well as a characterization of the nature and goals of diagnostic reasoning.

    Journal Title

    International Journal of Expert Systems

    Volume

    10

    Issue/Number

    2

    Publication Date

    1-1-1997

    Document Type

    Article

    Language

    English

    First Page

    177

    Last Page

    195

    WOS Identifier

    WOS:000071930600003

    ISSN

    0894-9077

    Share

    COinS