Database Reference
In-Depth Information
the reasoners HermiT,
9
FaCT++,
10
and Pellet,
11
and plug-ins for editing in the CNLs
Rabbit
12
(called the ROO tool) and AceView.
13
There is also an online version called
WebProtege.
14
Other editors are available, such as the open source NeOn-Toolkit,
15
which is a plug-in to the Java Integrated Development Environment Eclipse, the
domain-specific Snow Owl for clinical terminologies,
16
the online cloud-based
Knoodl,
17
and the commercial TopBraid Composer
18
and FluentEditor
19
(the latter is
CNL based). Which of these is selected for use will depend on your budget, project
size, and requirements; however, the main influencing factors will be
• Commercial versus open source: While the commercial editors' price tag
will come with product support and more robustness, there are strong com-
munities surrounding the main ontology editors Protégé and NeOn, so
questions can be answered on their forums, and there are a larger number
of useful plug-ins.
• RDFS versus OWL: If the main aim is to create a Linked Data set, a tool
that is limited to creating RDFS ontologies will be sufficient as priority
must be given to the management of large datasets. For more details of
such tools, see Chapter 7. Instead, if the aim of the project is to author
highly descriptive ontologies or to integrate ontologies together, one of the
ontology editing tools mentioned that support OWL 2 should be used.
•
Reasoning, rules, and queries.
The OWL 2 reasoners FaCT++, HermiT, and Pellet are all offered as stand-alone
or as plug-ins to Protégé 4. RacerPro
20
is a commercial reasoning system that is
available as a stand-alone product, with a visualization tool RacerPorter to manage
the knowledge bases. There are other reasoners that support subsets of OWL 2,
21
for example, ELK
22
and CEL,
23
which support OWL 2 EL; QuOnto, which supports
OWL QL; and Oracle 11g relational database management system, which supports
OWL RL. If you know your ontology conforms to one of the OWL profiles, it is gen-
erally better to choose a reasoner that corresponds to its logical complexity as they
are faster and lighter than the reasoners that cover the whole gamut of OWL 2 logic.
If you cannot decide, there is always TrOWL,
24
which is an interface to a number of
reasoners and offers EL (“TrOWL REL”) and QL (“TrOWL Quill”) or uses FaCT++,
HermiT, or Pellet for full DL reasoning. HermiT, Pellet, and RacerPro also provide
support for SWRL rules and allow SPARQL or SPARQL-DL (a subset of SPARQL
that supports OWL DL-based semantics) queries.
•
CNL support. As mentioned, Protégé has some plug-ins to enable authoring
in a CNL, for example, ROO and CloNE (Funk et al., 2007), and the com-
mercial tool Fluent Editor has integral support for its CNL. All others allow
annotation of the corresponding CNL statement using the
rdfs:comment
construct. If a domain expert, unfamiliar with OWL, is responsible for
authoring the ontology directly, then we would strongly advise considering
the use of an editing tool that supports CNL.
