 |
 |
 |
 |
|
| back | ||||
Abstract
An Approach to Large-Scale Pin-Map Problems
Track: Modeling
Author(s): Bart Pittari, Paul Crocker, Mitch Tinsley
This paper describes the rule development process and software issues encountered in the development of a geospatial expert system for a large (108 km2) area. The analysis included interviews and a neural network clustering analysis to develop rule hypotheses. The combination of these was then used to develop a set of rule hypotheses for statistical testing against random occurrence. Hypotheses testing significantly above or below random occurrence were used to produce GIS polygonal coverages that were combined using a weighted union to produce a cueing or coverage. This technique is felt to be applicable to a variety of problems.
Bart Pittari
University of Southern Mississippi
Center of Higher Learning
Bldg 1103 Room 103
Stennis Space Center , MS 39529
USA
Phone: 228-689-8125
Fax: 228-688-7454
E-mail: bart.pittari@usm.edu
Paul Crocker
University of Southern Mississippi
Center of Higher Learning
Bldg 1103 Room 103
Stennis Space Center 39529
USA
Phone: 228-689-8128
Fax: 228-688-7454
E-mail: paul.crocker@usm.edu
Mitch Tinsley
University of Southern Mississippi
Center of Higher Learning
Bldg 1103 Room 103
Stennis Space Center 39529
USA
Phone: 228-689-8187
Fax: 228-688-7454
E-mail: mitch.tinsley@usm.edu