(Description of Seminar 7)
Oct. 13, 1997
On last wednesday, Dr. Cobb gave us a lecture about A Knowledge-Based Approach from the Conflation of Vector Mapping Data. Dr. Cobb is a professor of Computer Science Department of USM (University of Southern Mississippi.
Dr. Cobb first explains what is conflation. She presents the cycle procedure from Feature Matching, to Feature Deconfliction, to Position Alignment, to Phase of Conflation, then back to Feature Matching. Then she discusses each step of this procedure.
For Feature Matching, pre-process steps to unify logical feature parts segmented during collection/production. Hierarchical rule base with matching criteria based on: 1. distance measure, 2. cartographic representation, 3. FACC feature and attribute codes, 4. neighborhood geometry, 5. topology. She also presents the user interaction.
Attribute Deconfliction must deconflict attribute sets and values. Attribute sets vary based on product guideline/use for matching. Attribute value deconfliction is rule-based processing involving information from data quality table/coverages, which includes source data, positional and attribute accuracy, logical consistancy, completeness and resolution.