SC 740 SEMINAR REVIEW
By Deborah Dent
Uncertainty Models in Geographic Information
Systems and Spatial Databases
Maria Cobb
Department of Computer Science and Statistics
University of Southern Mississippi
Hattiesburg, Mississippi
Friday, April 3, 1998
.Dr. Cobb presented a talk on the uncertainty models in Geographic Information System (GIS) for spatial databases. During her presentation she presented background information on GIS and spatial databases, focusing on ways in which various types of uncertainty are introduced in the process, together with a variety of fuzzy set approaches that have been developed to handle such uncertainty.
She began her talk by defining a GIS as a tool used for the analysis and storage of geographical data. Popular systems include:
ArcInfo
MapInfo
Typical uses include urban planning, pollution control, and crop planning.
Next she presented information on spatial data entry problems. The spatial information describes the location and shape of geographic features in terms of point, lines and areas. Spatial data modeling applications include CAD, VLSI and GIS. The main approach to examining the data is via queries. The two fundamental type of queries used are feature based and location based. She displayed a sample of a Spatial data integration which involves bringing together spatial data of different types (i.e. vector and raster) for a variety of information purposes, ranging from simple visual data overlays to automated data analysis. This was followed with examples of queries.
Dr. Cobb next discussed uncertainty. She pointed out that errors may occur in:
Data collection
Data classification
Data analysis
Perception of map content (visualization)
Representing reality
One of the main sources for uncertainty is cartographic license. The slogan used here is no data is worst than bad data. She then described the types of uncertainty and approaches for handling uncertainty. Two approaches discussed include:
Fuzzy Database Representation
Relational models
Dr. Cobb then presented material on dealing spatial data and introduced us to a variety of fuzzy set approaches. Dr. Cobb along with Dr. Petry of Tulane University will present a workshop at the FUZZ-IEEE 98 where they will present more detailed information on the subject matter.