Masters Theses

Date of Award

6-1985

Degree Type

Thesis

Degree Name

Master of Science

Major

Computer Science

Major Professor

J. Robin B. Cockett

Committee Members

Doyt Perry, Michael Thomason

Abstract

Decision trees are a form of knowledge representation that have lost popularity with the successes achieved by rule-based systems. One drawback to decision tree-based knowledge systems is their lack of hierarchical structure, which is essential for the implementation and maintenance of large scale knowledge systems (Woods). Also, explanations are difficult to generate because it is hard to focus the attention of a program on the problem at hand when a decision tree methodology is used.

Dr. J. Robin B. Cockett's recent work on decision theory and decision expressions (Cockett 1983a, 1983b, 1984) lead to the development, by the author and Dr. Cockett, of a decision expression interpreter called DECIDE. An augmented decision tree was used to represent knowledge compactly. Decision expressions were used to encode the knowledge in these augmented decision trees. The DECIDE language allows knowledge to be programmed hierarchically as knowledge clusters. Clustering provides a basis for both file handling and explanation generation. The DECIDE program manages a knowledge base of these clusters. It guides the user through the decision process to solve the user's problem. DECIDE also provides the user with a friendly and helpful environment in which to work.

Download
Files over 3MB may be slow to open. For best results, right-click and select "save as..."

Share

COinS