Database Reference
In-Depth Information
such as Spatial Relations contain only property definitions, not class definitions;
this is often a useful way to partition ontologies into modules.
10.3
METHODOLOGIES
There are a number of methodologies for developing ontologies. The benefit of using
a formal methodology is that it provides structure to the process and helps to ensure
best practice is carried out. We do not recommend a particular methodology but have
identified METHONTOLOGY (Fernández-López, Gómez-Pérez, and Jursito, 1997),
UPON (De Nicola, Missikoff, and Navigli, 2009), and Kanga (Mizen, Hart, and
Dolbear, 2005; Dennaux et al., 2012) as options. They are all fairly similar in their
approach, and UPON and Kanga have certainly borrowed from METHONTOLOGY.
Both UPON and Kanga add competency questions (Noy and McGuinness, 2001)
(questions that test that the ontology meets the purpose for which it is built); and
UPON also adds use cases. Whatever methodology you adopt, we would recommend
that at the very least the stages discussed next are used.
10.3.1
s
cope
anD
p
Urpose
One of the most important and often missed stages is the very first: defining what
the purpose of the ontology is and what its scope is. Without this, the construction of
the ontology can lose direction and focus. If we take the Spatial Relations ontology
as an example, we can say that the purpose is to provide a number of topological
relationships that can be used by anyone needing to include such relationships in
their applications and ontologies; we can say the scope is limited to RCC8 relations.
The Spatial Relations ontology is a small micro-ontology, and the scope and purpose
are quite easy to define. Defining the scope and purpose of domain ontologies can be
more challenging—especially the scope, as it is all too easy to try to include more
than is strictly necessary. Therefore, spending time up front thinking in detail about
the scope and purpose is time well invested. Doing so can bring substantial savings
in time later in the ontology authoring process, and indeed in helping to define an
RDF vocabulary as identified in Chapter 7, and in the creation and linking of data-
sets, covered in Chapter 8.
The scope and purpose will also help the authors to decide whether to use OWL
or RDFS to describe the ontology. There may be circumstances when this is not nec-
essary, most obviously if an organizational choice has mandated that all ontologies
will be authored using a specific language. Otherwise, the scope and purpose will
help to inform the choice of ontology language. The main criterion is how complex
(descriptive) the ontology needs to be, and put simply, if it does not require the com-
plexity of OWL, RDFS is a good choice.
10.3.2
U
se
c
ases
anD
c
ompetency
Q
Uestions
Use cases and competency questions are particularly useful for application ontolo-
gies that have specific uses. Use cases are simply descriptions of the uses to which
