Files in this item



application/pdf8409903.pdf (4MB)Restricted to U of Illinois
(no description provided)PDF


Title:Qualitative Reasoning in an Expert System Framework
Author(s):Cross, Stephen Edward
Department / Program:Electrical Engineering
Discipline:Electrical Engineering
Degree Granting Institution:University of Illinois at Urbana-Champaign
Subject(s):Engineering, Electronics and Electrical
Abstract:Expert systems typically utilize a declarative and uniform knowledge representation. The approach offers many operational advantages (e.g., a simple control structure), but is limited to expressing an expert's surface level knowledge in the form of pattern-decision pairs. The computer should have access to 'deeper' knowledge if it is to understand and justify its planning actions. Consider a domain where knowledge in the form of equations and algorithms is computationally too complex for use by the human practitioner. How should mathematical knowledge be represented to aid in the improvement and justification of plans? In the task domain of this research, enroute air traffic control, heuristically generated plans are justified by applying qualitative reasoning to aircraft performance equations. Equations are represented in a semantic network where nodes represent variables and links represent dependent variable influences.
The approach is unique in three aspects. First, a level of abstraction is included. Domain equations may be computationally too complex for a human expert to use. However, the equations can be interpreted in terms of a naive representation of Newton's laws as applied to one dimensional motion thus abstracting the influences inherent in the equations. Second, the approach enables bidirectional reasoning. Qualitative knowledge can be used to direct quantitative reasoning. Additionally, when new equations are implemented, their meaning is represented explicitly and interpreted using the existing qualitative knowledge. Third, the computer constructs its own representation of the equations based on a symbolic series expansion.
Issue Date:1983
Description:138 p.
Thesis (Ph.D.)--University of Illinois at Urbana-Champaign, 1983.
Other Identifier(s):(UMI)AAI8409903
Date Available in IDEALS:2014-12-15
Date Deposited:1983

This item appears in the following Collection(s)

Item Statistics