Abstract

During the last few years, expert system explanation has become an active research area after recognizing chat its role goes beyond expert system verification as assumed by early systems. After examining current systems, it has been found that only a few of these meet the requirements of an end user interested in learning more about the domain being addressed by an expert system and understanding why it has reached a certain conclusion. It was also found that systems that address these requirements, do so at a very high cost) since they embed unmense amounts of knowledge in a system without providing any means for accessing this knowledge except by the system for which it was built. The primary goal of this thesis is to investigate the use of an agent based approach for the explanation problem, such that knowledge re­usability would be promoted and high quality explanations generated. The motivation behind using an agent based approach follows from the realization that when addressing explanation on an end user level with the goal of ratification, an explanation system and an expert system, have separate goals. While one is concerned with strategies for clarifying and communicating with the user for explanation purposes, the other is primarily concerned with problem solving. Explanation on that level, requires deep knowledge and hence

nowledge representation forms and reasoning strategies that are different than those required by the expert system. On that level, the two systems are loosely coupled. However, the explanation system needs to "communicate" and "cooperate" with the expert system in order to carry out its own problem solving activities. This thesis describes the architecture within which a number of agents, each of which handles part of the explanation problem, should operate along side the expert system or other applications. Through the implementation of an experimental prototype, the approach presented was found to show great promise since it satisfied the addressed explanation goals, achieved knowledge re-usability, and modularity. The devised architecture was also found to be scalable & open, and to promote parallelism in the operations of the different agents.

School

School of Sciences and Engineering

Department

Computer Science & Engineering Department

Degree Name

MS in Computer Science

Date of Award

2-1-1998

Online Submission Date

1-1-1997

First Advisor

Ahmed S

Second Advisor

Ahmed Rafea

Committee Member 1

Adeeb El Ghonimy

Committee Member 2

Amr El Kadi

Document Type

Thesis

Extent

129 leaves

Library of Congress Subject Heading 1

Expert systems (Computer science)

Rights

The American University in Cairo grants authors of theses and dissertations a maximum embargo period of two years from the date of submission, upon request. After the embargo elapses, these documents are made available publicly. If you are the author of this thesis or dissertation, and would like to request an exceptional extension of the embargo period, please write to thesisadmin@aucegypt.edu

Call Number

Thesis 1997/58

Location

mgfth

Share

COinS