Title: GeoInterpret: A Framework for Ontology-Based Geoprocessing

When: Thursday, March 4th 2004, 9:30 am

Where: Room 503 IS Bldg.

Who: Ratchata Peachavanish

Advisor: Dr. Hassan Karimi, DIST, University of Pittsburgh

Committee:

Dr. Stephen Hirtle, DIST
Dr. Michael Lewis, DIST,
Dr. Jeen-Shang Lin, CE, University of Pittsburgh
Dr. Burcu, Akinci, CE, Carnegie Mellon University

Abstract: Today's GIS are sophisticated platforms for solving a wide range of problems across disciplines. However, despite the many benefits GIS platforms offer, their utilization requires knowledge and skills in understanding how real-world problems are modeled as geographic objects, determining geospatial data needs of problems, obtaining the required geospatial data sets, and deciding on the appropriate set of geoprocessing operations that is needed to solve a problem. In other words, users interested in utilizing GIS for solving application problems must gain an in-depth knowledge about how GIS model the real-world problems and what GIS techniques are needed to solve them. In doing so the most important phase of solving problems utilizing GIS is to map problems from application domains to the ways in which GIS can understand and solve them. Current technology in GIS does not have a means of such mapping and as a consequence, users are required to have an extensive training before they can effectively use GIS.

An ontology-based framework called GeoInterpret is proposed to improve this situation by allowing automatic translation of users' geospatial queries into geoprocessing workflows. The framework consists of an abstraction model that characterizes geospatial queries; a conceptualization of geospatial objects and geoprocessing operations; and a mechanism for workflow composition and workflow representation. The framework will be developed and validated by a proof-of-concept software program.