John Goodwin
Ordnance Survey

Ontology Construction in Ordnance Survey

(based on a paper being written by Glen Hart, Hayley Mizen and Sarah Temple)

“There is a theory which states that if ever anyone discovers what the universe is for and why it is here, it will instantly disappear and be replaced by something even more bizarre and inexplicable.

There is another which states that this has already happened”

The Restaurant at the End of the Universe - Douglas Adams

Introduction

Ordnance Survey of Great Britain is a national mapping agency in transition. Traditionally responsible for the production of mapping at scales between 1:250K and 1:1250 the organisation is now developing as a supplier of geographic information where this information may be delivered in the form of traditional mapping (paper and electronic), as discrete feature based digital products and through services. An already broad customer base is expanding from utilities, local and central government and the rambler, to encompass a wider range of commercial companies and the private user of the internet and mobile technologies.
The data that Ordnance Survey collects is being transformed from information suitable only to generate mapping to discrete feature based information suitable for use in a wide range of digital applications.

In parallel with the continual development and improvement of the information content that we hold there is a growing appreciation of the need to ensure this information is capable of being maximally interoperable with an end user’s own information. This appreciation has naturally led us to become curious about ontologies as a means to make our information more semantically meaningful. Our end goal therefore is to ensure that Ordnance Survey information can be understood and used by a third party system without human intervention. Whilst this will bring the enormous benefits of significantly reduced cost of adoption and minimises technical impedances we also appreciate that this is currently unobtainable given the current limits of computer science. So our lesser goal is to maximise what can be understood by machines.

Ontology Development

We are in the initial phase of an exploratory investigation into the development of ontologies and research into how they may be used enable information from heterogeneous sources to interoperate. Specifically, it describes our approach and initial experiences in constructing a subset of a full topographic ontology - one based largely around inland hydrology. The approach so far has been to consider the concepts we need to describe and the relationships between them. This was done intially at a philosophical level. We then tried to implement the ontology as best we could using existing Semantic Web Ontology languages such DAML+OIL and OWL. The work is as yet unfinished. We have still to finish refining the ontology and have only made a start in defining instance data based upon it. Some aspects will only become testable when we attempt to match it to a fresh water ecology ontology. Nonetheless useful knowledge has been gained.

The ontology we have produced is a network not a hierarchy with the taxonomic hierarchy being limited and disarticulated (there is no single root). We believe this has enabled us to introduce new concepts more easily. Certainly the existence of what hierarchies we have has always resulted in some debate when new concepts are presented. We believe this bodes well for the future where we may wish to share concept definitions across ontologies. There are clearly questions around whether DAML+OIL are sufficiently expressive enough to express what we believe we need. In part we are sure that there may exist some solutions within DAML+OIL that are inexperience has as yet failed to reveal. It also seems apparent that the rather fuzzy world of geography presents challenges that we are as yet unable to resolve and therefore that in the immediate future some compromises are required.