Ontologies for GIS
Investigator: Kathleen Stewart Hornsby
Ontologies provide a fundamental classification for geographic domains that captures, for example, categories of entities recognized for a domain, as well as the relations that link these categories. These relations include taxonomic relations such as isA, mereological relations including componentOf, and topological relations, for example, containedIn. In general, ontologies play an important role for knowledge representation, database design, information retrieval, and the semantic web, where they are used as an information engineering tool, for taxonomic reasoning and for first order logical inference. With respect to GIScience, ontologies have been promoted particularly for their role in reuse, sharing, and interoperability.
My research relating to ontologies begins with an examination of an ontology of change, where change in this case refers to identity change, i.e., the evolution of an entity over time including creation and elimination. Identity is distinct from an object's properties and allows one object to be distinguished from another. A set of basic identity changes are identified and these change primitives give the foundation for describing more complex changes. The kinds of change described with this approach reflect common types of change for many geographic objects such as water bodies that go in and out of existence or nations that are consumed through conflict or that are created new.
Basic symbols used for (a) object existence (b) non-existing object without history
and (c) non-existing object with history
Selected publications:
K. Hornsby (2001) Temporal zooming, Transactions in GIS. 5(3): 255-272.
K. Hornsby and M. Egenhofer (2000) Identity-based change: a foundation for spatio-temporal knowledge representation, International Journal of Geographical Information Science.14(3): 207-224.
K. Hornsby and M. Egenhofer (1998) "Identity-based change operations for composite objects," In: T. Poiker and N. Chrisman (Eds.) Proceedings of 8th International Symposium on Spatial Data Handling, Vancouver, Canada, pp. 202-213. pdf file
K. Hornsby and M. Egenhofer (1997) "Qualitative Representation of Change," In Spatial Information Theory - A Theoretical Basis for GIS, International Conference COSIT '97, Laurel Highlands, PA, Vol. 1329, Lecture Notes in Computer Science, S. Hirtle and A. Frank, Eds. Berlin: Springer-Verlag, 1997, pp. 15-33. pdf abstract
More recently, this research 
has expanded to consider the role of ontologies in dynamic geographic domains. For this work, we study event-based ontologies where the typical ontological classes are extended to describe not only continuant entities (e.g., houses, roads, airports) but also occurrents (e.g., snowstorms, traffic jams, plane taking off). Extending ontologies to incorporate events exposes new kinds of relations including object-event relations and event-event relations.
Selected publications
S. Hall and K. Hornsby (2005) Ordering events for dynamic geospatial domains, in A. Cohn and D. Mark (eds.) Proceedings of the Seventh International Conference on Spatial Information Theory, COSIT 2005, Lecture Notes in Computer Science 3693, Springer, Berlin, pp. 330-346.
K. Hornsby (2004) Retrieving event-based semantics from images, Proceedings of the IEEE Sixth International Symposium on Multimedia Software Engineering,13-15 December 2004, Miami, FL, IEEE Computer Society, pp. 529-536.
M. Worboys and K. Hornsby (2004) From objects to events: GEM, the geospatial event model, In M. Egenhofer, C. Freksa, H. Miller (Eds.) Proceeding of GIScience 2004, Lecture Notes in Computer Science, 3234, Springer, Berlin, pp. 327-343.
In recent research, we
present methods for automatically combining geospatial and temporal ontologies such that a geospatial domain can be analyzed over multiple temporal granularities. Terms from a geospatial ontology are combined with terms from a temporal ontology to form cross products that provide an integrated spatio-temporal reasoning framework. This framework is multi-granular, highlighting elements from the geospatial ontology at different times.
This new framework can be browsed, queried and visualized for more comprehensive analysis.
The terms in the cross product are a combination of geospatial and temporal terms and can be used, for example, for making higher-order inferences about a domain, such as retrieving the geospatial term CampusRoad over all possible times. In addition, the combination of terms can be used for automatically providing the complete set of possible annotations that can be used in conjunction with, for example, an image data set for a geospatial domain. We show how pairs of orthogonal ontologies represented in Protégé can be used as the input for deriving cross products and how the results of this technique can be used as a basis for reasoning about geospatial domains.
This research is supported in part by the National Geospatial-Intelligence Agency under grants NMA201-00-1-2009 and HM1582-05-1-2039.
