A Foundational Framework for Structuring Geographical Categories

Boyan Brodaric, Geological Survey of Canada Introduction Geographical categories are often represented in machine-readable ontologies to aid information interoperability, retrieval, fusion, analysis, workflow execution and image classification. Within these activities categories are routinely merged, split, and compared. Such operations are particularly dependent on the representation choices made at the syntactic, structural and semantic levels: e.g. category similarity can be determined according to syntactic resemblances of names, structural metrics such as distance in a graph, or semantic metrics such as possessing common relations. To enable these operations categories are regularly transformed to a common representation framework, suggesting a need for optimal strategies at each representation level. This paper builds on advances in foundational ontologies and geospatial semantics to develop a representation framework for geographical categories at the semantic level, using the Lake category in a running example.