Publication Date

9-2004

Committee Chair

Kelly Chan, Ph.D.

Committee Members

Kent R. Beaman, M.S.
Karen K. Kemp, Ph.D.

Abstract

Much useful data is currently not available for use in contemporary geographic information systems because location is provided as descriptive text and not in a recognized coordinate system format. This is particularly true for datasets with significant temporal depth such as museum collections. Development is just beginning on applications that automate the conversion of descriptive text based locations to geographic coordinate values. These applications are a type of geocoding or locator service and require functionality in two domains: natural language processing and geometric calculation. Natural language processing identifies the spatial semantics of the text describing a location and tags the individual text elements according to their spatially descriptive role. This is referred to as geoparsing. Once identified, these tagged text elements can be either converted directly to numeric values or used as pointers to geometric objects that represent geographic features identified in the description. These values and geometries can be employed in a series of functions to determine coordinates for the described location. This is referred to as geoprocessing.

Selection of appropriate text elements from a location description and ancillary data as input is critical for successful geocoding. The traverse, one of many types of location description is selected for geocoding development. Specific text elements with spatial meaning are identified and incorporated into an XML format for use as geoprocessing input. Information associated with the location is added to the XML format to maintain database relations and geoprocessing error checking functionality. ESRI’s ArcGIS 8.3 is used as a development environment where geoprocessing functionality is tested for XML elements using ArcObjects and VBA forms.

Share

Article Location

 
COinS