Extending Gazetteer Service in Geographic Information Retrieval A Position Paper

Hill (2000) identifies three basic elements in a digital gazetteer: name, footprint, and type (or category). Gazetteer entry types are organized in a hierarchical classification system. The name and footprint properties of geographical features enable automatic association between non-spatial data (e.g. text documents) and the location of geographical features, whereas the feature type hierarchy provides automatic organization of the the spatially indexed data. Used in a geographic information retrieval system (GIR, Larson 1996), such spatial indexing makes it possible to answer thematic (“what”) and locational (“where”) queries. The “theme ” in the query may be spatial, referring to a geographic feature (e.g. the City of Redlands) or to a category of such features (cities). It can also be non-spatial, referring to some property of the feature (e.g. the city’s population). While the name and footprint properties of geographic features are formally defined in a gazetteer, the typical set of properties associated with a particular feature category often are not. Similarly, the typical set of relationships among geographic features (other than the is-a relationship, such as the capital-of relationship) are often not present in gazetteers. GIR systems could potentially benefit from explicit definitions of these properties and relations by using them to automatically index and later retrieve relevant data, especially in the case of highly structured data such as GIS datasets. The Redlands Institute at the University of Redlands is conducting applied research in GIR using a prototyp