An ontology of spatial relations for the semantic web

Jerry Hobbs

Information Sciences Institute

University of Southern California

Marina del Rey, CA 90292

Abstract

I will describe an effort that is just beginning to develop a common ontology for expressing and reasoning about spatial information for the Semantic Web, as part of the DARPA Agent Markup Language (DAML) program. The aim of this ontology is to provide a way for different spatial reasoning engines and spatial resources to communicate with each other, as well as a way for people to mark up the spatial information on their web sites. The goals of the effort are to produce an ontology that will

1. Enable general, though not necessarily efficient, reasoning about spatial concepts.
2. Link with more efficient specialized reasoning engines for spatial reasoning.
3. Link with the numerous databases that exist containing a wealth of specific, e.g., geographical, spatial information.
4. Support convenient query capabilities for spatial information.

The topics we intend to cover include topological relations (e.g., RCC8), dimension, orientation, shape, measures like length, area and volume, latitude, longitude and elevation, and political subdivisions. Much of the work will be focused on geographical knowledge, but the intent is not to restrict ourselves to this domain alone. Topological spatial relations are important in microbiology, for example. Other application areas that have been suggested are the geology of earthquakes, NASA applications, computer graphics, and virtual reality.

Of course to do a thorough spatial ontology is an immense job, but we intend to limit the effort to linking with resources, rather than duplicating them. For example, the ontology should interface with a resource on the shapes of geographical regions, but not encode its internal representations.