Logo image
Path Understanding Using Geospatial Natural Language
Thesis   Open access

Path Understanding Using Geospatial Natural Language

Bradley Andrew Swain
University of West Florida Libraries
Master of Science (MS), University of West Florida
2009

Metrics

3 File views/ downloads
63 Record Views

Abstract

This thesis describes a novel path understanding system incorporating a unique method of gathering a geospatial corpus of driving path descriptions. The corpus starts as a collection of natural language descriptions of driving routes gathered in a series of short trips taken by a set of drivers, which is then annotated for use in path understanding. The developed system, described in this thesis, allows the corpus to be annotated with database objects representing locations (entities) being referenced in the path descriptions. The annotation tool, The gEoSpatial Language Annotator (TESLA) and the general principles and features which such an annotation system require are described. While the main impetus of this thesis is to describe the development and use of the annotation system, it also describes a general annotation approach and workflow for geospatial language corpora – a concept that is itself a novel area of study. Finally, the larger path understanding system into which the annotation tool fits is described briefly, along with a full account of how the various systems and subsystems work together. The thesis concludes with a discussion of the annotation results and potential uses of the system, as well as opportunities for future work.
pdf
Path Understanding Using Geospatial Natural Language940.50 kBDownloadView
Preprint Thesis pdf Open Access

Details

Logo image