Database Reference
In-Depth Information
10.2.2 a
pplication
o
ntologies
Application ontologies differ from domain ontologies as they include references to
terms specific to a particular application or task. Domain ontologies will be used by
application ontologies to draw on terms common to both the domain and the applica-
tion; the application ontology will then add further terms that are related specifically
to the application. We can imagine Merea Nature having an application to monitor
the health of various habitats. It will construct the application ontology by taking
elements of the topographic domain ontology constructed by Merea Maps, per-
haps also a species definition from a domain ontology produced by an international
wildlife organization, and its own domain ontology defining the habitats found on
Merea. To this they will add terms specific to the application, possibly related to the
monitoring of the habitats or events such as fire, drought, or flood that could affect
the quality of habitats. The end result is something with a very specific purpose, less
likely to be reusable by others than a domain ontology.
10.2.3 t
op
-l
evel
o
ntologies
or
U
pper
o
ntologies
Top-level ontologies provide general vocabularies that can be utilized by domain and
application ontologies. Top-level ontologies were some of the earliest to be developed
and either tried to be encyclopedic in their coverage, such as OpenCyc,
1
or abstract
in the extreme to generalize every entity into atomic concepts, such as SUMO
2
and
DOLCE
3
; as a result, they are very large and unwieldy. The problem with these
ontologies is that human knowledge is founded within context; these ontologies are
not specialized and can become too philosophical. They use quite academic termi-
nology like “endurant” and “perdurant” that are not in common usage and can be
difficult to understand. It can become a challenge for the domain ontology author to
continually try to fit the concepts in the domain ontology into the structure of the
upper ontology. However, these ontologies do represent a considerable amount of
thought and so will contain well-formed solutions to specific areas or knowledge
modeling patterns, and these may well be worth reusing even if you do not want to
use the entire ontology.
10.2.4 m
icro
-
ontologies
Micro-ontologies are a more recent attempt to develop authoritative and reusable
ontologies. These ontologies attempt to provide very specialized terms specific
to well-defined domains and expertise. Examples are Dublin Core (International
Organization for Standardization [ISO], 15836:2009) and the Spatial Relations ontol-
ogy within GeoSPARQL (Perry and Herring, 2011). The first contains well-defined
terms for describing metadata and the latter specific spatial relationships related to
RCC8 (Region Connection Calculus 8). These are far more usable than top-level
ontologies since they provide useful terms in small packages. Developing such
micro-ontologies is generally good practice as it maximizes the chance of their reuse
by others (the larger the ontology, the more likely you are to disagree with some
terms within it even if you agree with others). It is also worth noting that ontologies
