Date of Award

5-2015

Level of Access Assigned by Author

Open-Access Thesis

Degree Name

Master of Science (MS)

Department

Spatial Information Science and Engineering

Advisor

Kate Beard-Tisdale

Second Committee Member

Max Egenhofer

Third Committee Member

Torsten Hahmann

Abstract

With the increased availability of geospatial data and efficient geo-referencing services, people are now more likely to engage in geospatial searches for information on the Web. Searching by address is supported by geocoding which converts an address to a geographic coordinate. Addresses are one form of geospatial referencing that are relatively well understood and easy for people to use, but place names are generally the most intuitive natural language expressions that people use for locations. This thesis presents an approach, for enhancing place name searches with a geo-ontology and a semantically enabled gazetteer. This approach investigates the extension of general spatial relationships to domain specific semantically rich concepts and spatial relationships. Hydrography is selected as the domain, and the thesis investigates the specification of semantic relationships between hydrographic features as functions of spatial relationships between their footprints.

A Gazetteer Ontology (GazOntology) based on ISO Standards is developed to associate a feature with a Spatial Reference. The Spatial Reference can be a GeoIdentifier which is a text based representation of a feature usually a place name or zip code or the spatial reference can be a Geometry representation which is a spatial footprint of the feature. A Hydrological Features Ontology (HydroOntology) is developed to model canonical forms of hydrological features and their hydrological relationships. The classes modelled are endurant classes modelled in foundational ontologies such as DOLCE. Semantics of these relationships in a hydrological context are specified in a HydroOntology.

The HydroOntology and GazOntology can be viewed as the semantic schema for the HydroGazetteer. The HydroGazetteer was developed as an RDF triplestore and populated with instances of named hydrographic features from the National Hydrography Dataset (NHD) for several watersheds in the state of Maine. In order to determine what instances of surface hydrology features participate in the specified semantic relationships, information was obtained through spatial analysis of the National Hydrography Dataset (NHD), the NHDPlus data set and the Geographic Names Information System (GNIS). The 9 intersection model between point, line, directed line, and region geometries which identifies sets of relationship between geometries independent of what these geometries represent in the world provided the basis for identifying semantic relationships between the canonical hydrographic feature types.

The developed ontologies enable the HydroGazetteer to answer different categories of queries, namely place name queries involving the taxonomy of feature types, queries on relations between named places, and place name queries with reasoning. A simple user interface to select a hydrological relationship and a hydrological feature name was developed and the results are displayed on a USGS topographic base map. The approach demonstrates that spatial semantics can provide effective query disambiguation and more targeted spatial queries between named places based on relationships such as upstream, downstream, or flows through.

Share