Petri net applications in the design and analysis of expert systems.
Item
-
Title
-
Petri net applications in the design and analysis of expert systems.
-
Identifier
-
AAI9000671
-
identifier
-
9000671
-
Creator
-
Bassen, Gordon Stuart.
-
Contributor
-
Adviser: Ken McAloon
-
Date
-
1989
-
Language
-
English
-
Publisher
-
City University of New York.
-
Subject
-
Computer Science
-
Abstract
-
This work presents the concept and implementation of a Rule-Based Petri Net (RBPN) class, as an addition to the family of Petri Net (PN) model extensions. We demonstrate that these RBPN's are capable of modeling a generic type of expert system shell. We further show how they may function as a tool in the design of expert systems; we also discuss their usefulness in the analysis of a developing expert system. As a design tool, we present RBPN concepts in the development of Production Rule systems. We also indicate the added RBPN contributions toward Conflict Resolution techniques for all types of Rule-Based systems. Before describing our RBPN, we discuss several of the extensions to the basic PN model that were precursors to the formal development of the RBPN.;Our RBPN has been developed with the following characteristics: two transition types, the first representing a rule statement, and the second a query transition for determining qualifier existence; two place types, one containing standard PN-type (simple) tokens, and the second place type allowing for data to be represented as bags of information. These data places are further refined to allow for choice, qualifier (both fixed and variable), and conditional place types.;In addition, data places do not have their values exhausted by transition firings, as is the usual case with simple place types in basic Petri Net models. This is done to maintain the attribute value(s) in a qualifier place unless the values are otherwise modified as a result of a transition firing.;We then illustrate the use of our RBPN for modeling a small expert system for mathematics remediation placement. This system allows us to show the capabilities of the RBPN as an aid in the design and development of expert systems. The small remediation system model is followed by a larger, more extensive expert system for student advisement of Data Processing majors at Kingsborough Community College. This system includes advisement for both the major area courses in Data Processing and the general academic requirements. In the future we hope to extend the system for advisement in all areas of study.
-
Type
-
dissertation
-
Source
-
PQT Legacy CUNY.xlsx
-
degree
-
Ph.D.
