Database Reference
In-Depth Information
9.1.2 s
yntaxes
Since an OWL ontology is an RDF graph, it can be written using many different syn-
taxes. We have already met the RDF/XML syntax in Chapter 5, and the RDF constructs
we are familiar with also form part of the RDF/XML syntax for OWL. In addition,
there are tags such as
owl:Class
that are used as part of the OWL RDF/XML syn-
tax. This is the only syntax that all OWL tools are required to support, so you are likely
to come across OWL ontologies formatted this way.
The Manchester Syntax (Horridge and Patel-Schneider, 2009) was introduced in
the OWL 2 standard with the aim of making it easier to read and write ontologies.
It is more compact and easier to understand than RDF/XML, so we use it in our
OWL examples in this topic.
For completeness, we also mention the OWL/XML syntax (Motik, Parsia, and
Patel-Schneider, 2009), which is easier to parse than RDF/XML and unlike the latter,
can be processed and queried using off-the-shelf XML tools like XSLT and XQuery.
However, like RDF/XML, it is still quite verbose, so we stick to the Manchester
Syntax for the rest of this chapter.
Although not part of the formal OWL standard, several controlled natural lan-
guage (CNL) syntaxes have been proposed for authoring OWL ontologies. These
include Attempto Controlled English (ACE) (Kaljurand and Fuchs, 2007); the
Sydney OWL Syntax (Cregan, Schwitter, and Meyer, 2007); and Rabbit (Hart,
Johnson, and Dolbear, 2008). Their purpose is to make it easier for experts in the
particular domain of knowledge—flood risk management, oncology, patent law,
or whatever it may be—to capture their own knowledge as easily and naturally as
possible, using sentences that are close to normal English, while maintaining the
unambiguity necessitated by OWL. The main differences between these three is that
ACE and the Sydney OWL Syntax both started life as more complex controlled lan-
guages, based on first-order logic (the latter as the language PENG), and were later
reduced in scope to conform to OWL DL, while Rabbit was developed by domain
experts looking to author, and understand, their own ontologies. In this topic, we
include Rabbit sentences as an explanation of the OWL DL examples. Appendix B
summarizes the primary constructs of Rabbit and their corresponding OWL DL
axioms in Manchester and OWL/XML syntaxes.
9.3
OWL LANGUAGE ELEMENTS
9.3.1
O
ntOLOgy
-L
eveL
c
Onstructs
An OWL ontology consists of a set of statements known as “axioms” that are usually
preceded with a set of namespace declarations. As with RDFS, these axioms consist
of triples with the structure <subject> <predicate> <object>, and like RDFS, OWL
implements these through classes, properties, individuals, and values.
The namespace declarations are there to identify the ontologies that are imported
and to denote the current ontology's own prefix, which is then used in the rest of the
ontology as a shortcut. It can be useful to introduce this prefix so that if the domain
