Jayant Sharma

Date of Award


Level of Access Assigned by Author

Open-Access Dissertation

Degree Name

Doctor of Philosophy (PhD)


Spatial Information Science and Engineering


Max J. Egenhofer

Second Committee Member

M. Kate Beard-Tisdale

Third Committee Member

Andrew U. Frank


As Geographic Information Systems (GISs) have gained wider acceptance and application, the interest in spatial modeling and reasoning has deepened. The purpose of a model and reasoning framework is to permit a user to experiment and study causal relationships in the real world by operating on the model only. Traditionally GISs employ purely quantitative methods to represent and infer spatial information. This approach has serious shortcomings when dealing with qualitative spatial information, which may be incomplete or imprecise and does not contain knowledge of the geometry of the spatial objects, which humans are particularly adept at utilizing in their daily lives. A crucial part of this human commonsense reasoning is the representation and manipulation of knowledge about time and space. This thesis is concerned with the construction of a qualitative spatial reasoner which mimics human commonsense reasoning about spatial relations. The spatial relations considered are topological relations, cardinal directions, and qualitative distances. The purpose of the integrated qualitative spatial reasoner is to allow inferences on individual or combined knowledge of spatial relations. Homogeneous spatial reasoning, for example, the inference of new topological relations from a set of topological relations, is achieved by using relation algebras developed for individual spatial relations such as cardinal directions or topological relations. Heterogeneous spatial reasoning, which allows the inference of relations A North of C and A disjoint C from the facts A North of B and B contain C, is made possible by using a canonical representation. A prototype implementation of the models proposed in this thesis demonstrates that (i) using individual spatial relastion algebras and an object-oriented design methodology facilitates the construction of a powerful modular and extensible intergrated reasoning subsystem, and (ii) using a canonical representation leads to a powerful heterogeneous reasoning mechanism that enables complex three-step inferences over topological and directions relations, that are not possible with conventional techniques for compositions of spatial relations, such as A West of B, B overlap C, and C West of D, implies A West of D.