Decision-Analytic Approach to Rule-Based Expert System Development Using GPS (Global Positioning System) as the Model
Abstract
The applicability of using a decision theory approach towards reasoning in expert systems is presented. Particular emphasis is placed on the management of uncertainty and how it can be consistently factored into decision making. Using a PC-based Scheme program as a decision-analytic reasoner, and a Quattro generated file composed of user provided likelihoods as the knowledge base, this capability is demonstrated using an anomalous condition onboard the GPS satellite as the test scenario. The model, using Jeffrey's Rule as a manner of manipulating uncertainty, is able to effectively capture GPS system knowledge probabilistically in a manner that matches expectations of the experts as well as coincides with the GPS Orbital Operations Handbook. How rules are represented in this type of system is discussed as well as the assumptions which go into the model's application. Limitations of the system and recommendations for expanding its uses are also given. Theses. (rrh)
Document Details
- Document Type
- Technical Report
- Publication Date
- Dec 01, 1989
- Accession Number
- ADA216179
Entities
People
- Grady N. Elliott Jr
Organizations
- Air Force Institute of Technology