Scientists and engineers using geographic databases need query languages with powerful spatial selection methods and capabilities to infer spatial information in a manner similar to a human expert. Crucial for geographic databases, containing very large amounts of spatial data, are appropriate operators to access and manipulate spatial data in large-scale, geographic space, far beyond what is currently being offered by traditional database management systems. The objective of the investigations is to construct a coherent reasoning system that integrates spatial concepts about topology, cardinal directions, and approximate distances so that they can serve as a spatial extension to geographic databases and query languages. The reasoning system will focus on large-scale, geographic space. The hypothesis is that powerful and complex spatial reasoning can be formalized as the product of the interaction between relatively simple spatial relations with specific inference rules. First, individual spatial relations of topology, distance, and direction will be formalized. Afterwards these formalizations will be used to develop an integrated, comprehensive system, i.e., a more complex algebra, from the previously defined ones, adding more power through the coexistence of the different relationships in a single system. Finally, the reasoning rules will be incorporated into a query processor that integrates qualitative and quantitative spatial information. The prototype system will be tested for ad hoc site selection in emergency situations with data from the Digital Chart of the World. The major result will be a set of primitives, with rules describing their combinations, for the design of domain specific spatial query languages.
Last updated on August 17, 1995.
[ Spatial Reasoning | Spatial Database Research Group | NCGIA Maine ]