Abstract: An object-oriented conceptual model for geographic annotation

By Tom Elliott for Terrain Summit 07 (Huntsville: 24-25 September 2007).

At a recent workshop on "Digital Gazetteer Research and Practice," convened at the University of California Santa Barbara under the auspices of the National Geospatial Intelligence Agency, conferees were unable to arrive at a universal definition of a digital gazetteer. This difficulty may be attributable, in part, to the diversity of the participants, a group that included intelligence professionals, academic geographers, commercial cartographers, historians, publishers, GIS analysts, computer scientists and representatives of the "location services" industry. They were able, however, to agree that "gazetteers" in the broadest sense constitute no more (and no less) than the "annotation layer" for geospatial data of all types. Names, feature classifications, textual descriptions and temporal information -- together with links to external resources -- are all forms of annotation frequently required in a variety of research, training and field contexts.

Another point of agreement was the idea of "digital gazetteers" as systems of knowledge organization (or conceptual models) that offer distinct advantages over traditional GIS data formats in some applications. In particular, the simple, flat GIS attribute table fails to support spatial applications that require annotation on the basis of heterogeneous, sparse, damaged or poorly understood data sources and documents. It is possible to construct relational databases that accommodate these needs; indeed, many are in production today. Nonetheless, these systems tend to the idiosyncratic, and the data they contain are often difficult to serialize into standard formats for interchange and archiving. The introduction of flags for quality and certainty in the parsing and assignment of annotation can further complicate such models.

The Ancient World Mapping Center's Pleiades Project, with startup funding from the National Endowment for the Humanities, has developed a novel object-oriented model for annotated geographic information. This paper will present the model in theoretical terms, and demonstrate its realization in the project's open source, web-based content management system. The presentation will illustrate the model's 5 main classes (names, locations, places, events and references) with special attention to support for indicating accuracy, certainty, completeness, language/script and other measures with respect to class attributes and the relationships between objects. The paper will also consider the current state of file formats for the storage and interchange of data organized according to this model, as well as model application in fields beyond historical geography.