Mereotopology for geospatial semantic integration

The current approaches of geospatial semantic integration mainly rely on standards, distinguishing features, ontology, communication procedure, and hierarchical structure to exchange geospatial data semantics and thus integrate the data. These research efforts require the infusion of extra information and/or tend to be slow and expensive. Furthermore, because the literature ignores spatial information, it overlooks some essential geospatial semantics and also wastes a great deal of geospatial data.; This dissertation proposes an innovative approach to applying implicit non-spatial (linguistic) and derived spatial information in geospatial semantic integration. This dissertation enhances the theories and methodologies of mereology and mereotopology that seek to improve and automate semantic definitions. The mereology approach uses subcategories and descriptions as implicit non-spatial information to automate semantic definition, and the spatial pattern mereotopology incorporates derived spatial patterns to improve semantic definition. The mereology approach develops a Matching-Count Estimate algorithm for the Naive Bayesian Classifier to measure semantic similarity with a full consideration of the uniqueness of categorical geospatial data. The mereotopology method is based on the validation of a spatial neighborhood pattern assumption, and an algorithm of graph matching is developed to measure semantic similarity. This dissertation concludes that, by producing results comparable in accuracy to the results presented in the literature, the mereology and mereotopology methods can improve current semantic integration research with their low cost, high speed, and medium-to-high accuracy.; This dissertation proposes a solution to a key emerging research question concerning semantic integration in Geographic Information Science. It also aims to contribute to the future development of Geographic Information System, in particular, to distributed Web-based GIS, and fits the research of spatial ontology in with the theory and methodology of mereotopology. Finally, this study involves interdisciplinary theories and methodologies of geography, cognitive science, databases, mathematics, and land use.