Database Reference
In-Depth Information
specify the “greater than” aspect, so what it does is create a value “LakeSize” and
property hasSize that is used to denominate any pond, reservoir, and lake that is
greater than 2 hectares. This value is assigned as a part of the preprocessing that
Merean Nature does when importing the topographical data it receives from Merea
Maps. The Merean Nature's Lake class can then be made equivalent to an anony-
mous class (Pond or Reservoir or Lake) and hasSize LakeSize:
Class: Lake
EquivalentTo: (mm:Pond or mm:Reservoir or mm:Lake) that hasSize value
LakeSize 22
As Merean Nature is preprocessing the data, it could of course simply classify
every eligible feature from the Merea Maps' dataset directly as a Merean Nature
Lake and not worry about the ontological solution that has just been presented.
On the face of it, this seems a reasonable solution and will certainly work. However,
the downside is that this hides the relationship in code rather than making it explicit,
which is one of the major points of using an ontology.
There will of course be areas where nothing can be done, and no relationship can
be reasonably constructed between two classes, for example, if the external dataset
lacks the data needed to extract such a relationship and its ontology can provide no
corresponding description. As a case in point, Merean Nature has a habitat class Still
Anoxic Freshwater. Such a habitat can occur in things that Merea Maps has classified
as Pools, Lakes, Reservoirs, Canals, and Rivers and Streams, but only where there
is no low and the oxygen levels are very low. Merea Maps simply does not record
oxygen levels, and still waters can occur in parts of all these freshwater features, but
again other than for Ponds, Merea Maps does not record where still waters occur.
So, the reality is that there is little that Merean Nature can do in this situation as there
is no useful Merea Maps class to which they can link.
One final thing to consider is: Where class equivalence is known to be valid, is
it better to create a local class and make it equivalent to the external class, or is it
better to just use the external class? If an area as young as ontology authoring can be
said to have a tradition, then it has been traditional to say that the preferred solution
is always to reuse the external class rather than also creating a local class. However,
there is a downside to this because more and more ontologies need to be referenced
the further down the reuse chain you go, and the system can become unmanage-
able. So, there is an argument to say that if you in turn expect your ontology to be
reused by others, then it may be more sensible to create local classes and make them
equivalent in a separate ontology file so that others who wish to reuse your ontology,
but are not necessarily interested in other ontologies, can do so without pulling in
too many unnecessary ontologies. Fig u re  10.11 shows how this works. The Main
Local Ontology contains the body of the ontology, including local descriptions for
the classes that are also present in the External Source Ontology. In  the case of
Merean Nature, the Main Local Ontology would contain all its habitat descriptions
along with local classes for things such as Woodland. The Merea Maps ontology
is the External Source Ontology and also contains a description of Woodland.
These classes are linked by a separate Merean Nature ontology that just contains
Search WWH ::




Custom Search