Database Reference
In-Depth Information
10
Building Geographic
Ontologies
10.1 INTRODUCTION
In Chapters 6 and 7, we saw how Geographic Information (GI) can be represented as
Linked Data using RDF (Resource Description Framework) and described at a basic
level using RDFS (RDF Schema). This chapter builds on that baseline by showing
how Linked Data can be more richly described using ontologies written in OWL
(Web Ontology Language). It provides an overview of a method to build ontologies
and then discusses the issues and techniques of ontology building with respect to GI.
The techniques and problems dealt with are not exclusive to geography, but geogra-
phy is distinguished by the degree and frequency to which certain issues occur. The
techniques and approaches described here are therefore applicable in many other
domains. The examples are expressed in the Rabbit controlled natural language,
followed by the same information in OWL, using the Manchester syntax.
The OWL language itself was described in Chapter 9, and while this current
chapter gives examples using many of the features described in Chapter 9, it does
not present exhaustive examples of all the language features. This is because the
emphasis in this chapter is to demonstrate how certain key characteristics of GI can
be ontologically described, rather than attempting to present geographic examples
of each OWL construct.
10.2 TYPES OF ONTOLOGY
Ontologies can be characterized in different ways, in overlapping categories, such
as top level, domain, application, micro-ontologies, and link ontologies. While
we have already mentioned link ontologies in association with link generation in
Chapter 8, here we mainly deal with domain ontologies but also touch on top-level
and micro-ontologies.
10.2.1 D
omain
o
ntologies
A domain ontology is one that describes the vocabulary and relationships that are
associated with a particular domain of interest. It is application independent and will
not contain terms specific to a particular application; rather, it will contain descrip-
tions for the terms used in a general area of expertise. Organizations that supply gen-
eral application-independent data, such as Merea Maps, are likely to produce domain
ontologies. In the case of Merea Maps, they will be interested in developing a domain
ontology that describes all the topographic things that are represented on their maps.
183
