Handling vagueness in ontologies of geographical information

This thesis presents a novel approach to the problem of handling vagueness in ontologies of geographical information. Ontologies are formal representations of a set of concepts and the relationships that hold between those concepts. They have been proposed as a method of representing geographical information logically, but existing limitations in ontology languages and approaches fail to handle aspects of the geographical domain adequately, such as vagueness. The technique introduced in this thesis does not seek to remove or ignore the inherent vagueness when reasoning about geographic features, but rather seeks to incorporate it into decisions made about features during this process. By improving themanner in which vagueness is handled in geographical information systems, we improve the usability and the functionality of such systems, and move towards a more natural method of interaction. A comparison of the principal vague reasoning approaches is presented, to show how there is not at present a universal approach that handles all forms of vagueness. Rather, there exist different forms of vagueness as well as different required outcomes of vague reasoning, which means we should instead consider the problem at hand and determine the most appropriate approach accordingly. The technique for handling vagueness proposed here is to provide a systemfor grounding an ontology upon a geographic dataset. This data is assumed to take the form of a set of 2-dimensional polygons, each of which may be associated with one or more labels describing the type of region that polygon represents and the attributes associated with it. By grounding the ontology onto the data, an explicit link is made between the ontology and the data. Thus, vagueness within the definitions at the ontology level can be handled within the context of the dataset used; “large” can be defined in terms of what it means to be “large” in this dataset. Further, I developed a system that allows querying of the data and returns features through spatial reasoning. This allows the extraction

Name dc:type.qualificationname

Ph.D

Level dc:type.qualificationlevel

doctoral

Grantor dc:publisher.institution

University of Leeds

Year dc:date.issued

2008

Author dc:creator

• Mallenby, David

Advisors dc:contributor.advisor

• Cohn, A.G.
• Bennett, B.

Identifier

uk.bl.ethos.494607

OAI identifier oai:identifier

oai:etheses.whiterose.ac.uk:1373