Open Access
Soon, Kean Huat
Graduate Program:
Master of Science
Document Type:
Master Thesis
Date of Defense:
Committee Members:
  • Douglas A M Iller, Thesis Advisor
  • Douglas Alan Miller, Thesis Advisor
  • evolution
  • ontology development
  • matching gazetteers
  • geographic ontology
  • conceptual neighborhoods
In applications such as web-based information retrieval, gazetteers are often used to provide information about places in the real world, including names, geographic coordinates, and feature class membership. Although this information is often sufficient to differentiate places with ambiguous geographic names for a given point in time, the information contained in standard gazetteers does not account for the dynamic aspect of geographic features that evolve over time. When utilizing gazetteers that describe geographic entities this shortcoming can cause two different places to be considered identical, or the same place to be viewed as different entities. Misidentifying places may cause significant mistakes in decision-making during crisis management or national security situations by missing critical information associated with a place of interest or by incorporating erroneous information. To support the temporal continuity of geographical features, this thesis develops a geographic ontology that provides information regarding the evolution of coastal geographic features. This problem context is chosen because the changes in such features are strongly influenced by well-known earth system processes, such as sea level rise. Grounded in the notion of low-resolution conceptual neighborhoods, this thesis combines spatial data with the GeoNames gazetteer to develop the geographic ontology of coastal features. This ontology extends standard approaches by making explicit the knowledge of evolution of the spatial configuration of geographic features and their topological relationships. Using two local datasets from Florida and Alaska, the evaluation has shown that this type of geographic ontology can improve the automated matching between place names. This result in turn supports automated utilization of gazetteer-based information that represents the land surface at different times.