56 ◾ Rajendra Akerkar
<owl:Class rdf:about=”#Hotel”/>
</rdf:type>
</owl:Individual>
An axiom is dened as a class, property, or individual. Axioms are used to associ-
ate class and property identiers with partial or complete specications of their
characteristics and reveal other logical information about classes and properties.
Figure2.7 illustrates the language construct using OWLviz. e three OWL lan-
guage constructs for class axioms are rdfs:subClassOf, owl:equivalentClass, and
owl:disjointWith. In the following example (see gure), we present disjoint classes
using the owl:disjointWith property; it shows that DatePeriod and TimePeriod
have no instances in common.
<owl:Class rdf:ID=”DatePeriod”>
<owl:disjointWith rdf:resource=”#TimePeriod”>
</owl:Class>
ere is a dierence in the owl:disjointWith and owl:complementOf properties.
Using owl:complementOf, we can infer that period is not a date; it is a time. We
cannot make the same inference using the owl:disjointWith property. In another
example, we specify the rdfs:subClassOf property. is class is a concrete represen-
tation of the concept of seasons
<owl:Class rdf:ID=”Season”>
<rdfs:subClassOf>
<owl:Class rdf:ID=”DateTime”/>
</rdfs:subClassOf>
</owl:Class>
ing
is-a
is-a
Period
OpeningHours
Season
DatePeriod
DateTime
TimePeriod
is-a
is-a
is-a
is-a
Figure 2.7 Language construct using OWLviz.
Ontology: Fundamentals and Languages ◾ 57
2.8.4 OWL 2 Profiles
OWL 2 provides three proles: OWL 2 EL, OWL 2 QL, and OWL 2 RL. Each
prole targets dierent application scenarios. OWL 2 proles are dened by plac-
ing restrictions on the functional style syntax of OWL 2. An ontology written in
any of these proles is a valid OWL 2 ontology. Each prole is designed to trade
some expressive power for eciency of reasoning. For example, the OWL 2 EL
prole trades expressivity for the benet of polynomial time subsumption testing.
Similarly, reasoning for the OWL 2 RL prole can be implemented using a rule
engine. e important modeling features of OWL 2 EL are class conjunction and
SomeValuesFrom restrictions. e use of negation, disjunction, AllValuesFrom
restrictions, and cardinality restrictions are not allowed to achieve tractability. To
preserve its good computational properties, the datatypes supported by OWL 2
EL were selected to ensure that their intersection is either empty or innite. Many
large-scale ontologies may be captured using this prole.
OWL 2 QL is basically designed for data-driven applications and provides a
suitable means for RDF vendors to include some kind of OWL support without
excluding the advantages of a database type implementation. OWL 2 QL is based
on the DL Lite family of description logics. Its special quality is the inclusion of
many features required for capturing conceptual models. is prole is suitable
when a limited extension of RDF schema is desired.
OWL 2 RL was designed to allow the smooth adoption of OWL by vendors of
rule-based inference tools. Naturally, it oers better interoperability with knowl-
edge representation languages based on rules. It allows for most constructs of OWL
2, but to permit rule-based implementations of reasoning, the use of these con-
structs in axioms is restricted.
2.9 Integration of Ontology and Rule Languages
Integrating Semantic Web ontology with logic programming and business rules has
proven a challenge. According to Brachman et al. (1990), attempts to combine rules
with description logics were made in the days of the classic system. Semantic Web
Rule Language (SWRL) [http://www.w3.org/Submission/SWRL/] is a step in this
direction. SWRL was rst named OWL Rules Language before gaining built-in
atoms and a more RuleML [http://www.ruleml.org/] avored syntax. In that form,
it was published as a W3C member submission. e decidability of SWRL rules
with the DL safety condition was established by Motik et al. in 2004 and 2005 and
further elaborated by Motik (2006).
Another tool named Description Logic Program (DLP) inherits its seman-
tics from OWL and may be transformed syntactically into Logic Programming
syntax, thus providing a kind of basic interoperability between OWL and Logic
Programming. According to an analysis by Volz (2004), most existing OWL
58 ◾ Rajendra Akerkar
ontologies are almost completely contained in DLP. Another example is the
Semantic Web Research Community (SWRC) ontology (Sure et al., 2005). DLP
was originally presented in the works of Grosof et al. (2003) and Volz (2004).
Hybrid systems comprising both classical OWL reasoning and traditional rule-
based approaches like logic programming in dierent variants present concerns.
One hybrid solution is the MKNF knowledge base discussed by Motik et al. (2006;
2007). Another approach is based on integration of OWL DL reasoning with
Answer Set programming through the dlvhex system (Eiter et al., 2006b and c;
Schindlauer, 2006). According to Hitzler and Parsia (2009), such integration is not
as strong as hybrid MKNF knowledge bases and basically consists of two reasoning
engines that interact bidirectionally when reasoning over knowledge bases.
2.10 Ontology-Driven Information Integration
Data sets of interest to computational biologists are often heterogeneous in struc-
ture, content, and semantics. Such data sources are large, diverse in structure and
content, typically autonomously maintained, and need integration before utiliza-
tion. e next generation of computer-based information requires capabilities to
deal with such heterogeneous data sources available in distributed fashion, i.e., on
the Web. Ontology plays a key role. Ontology-driven information systems (ODIS)
are based on explicit use of technologies for computer-based information systems
(Guarino, 1998). As stated earlier, a software specication role for ontology was
suggested by Gruber (1991).
It is obvious that ontology can be generated using dierent representation lan-
guages based on various knowledge representation paradigms (description logics,
frame logics, etc.). According to Yildiz and Miksch (2007), to reduce the integra-
tion and run-time costs of ontology, ontology engineering should be automated to
a large extent and ontology management services must be provided in form of an
ontology management module (OMM). Information integration incorporates three
phases: (1) ontology generation, (2) ontology management, and (3) ontology inte-
gration. Ontology and information system integration remain challenges because
of the nature of ontology. Ontologies were conceived as backbones of semantic
networks to represent content eciently on the Web. Because information systems
do not share the characteristics of the Semantic Web, it is dicult to add Semantic
Web ontology to information systems.
Yildiz and Miksch (2007) cited requirements for fostering the wide acceptance
of ontology-driven information system development. ese requirements constitute
an abstract ontology model for representing additional semantic knowledge and for
ontology integration: (1) evolutional properties to indicate the expected behaviors
of particular components over time; (2) quality properties to indicate condence
levels of ontology components; and (3) temporal properties to mark transaction
times and valid times of components.
Ontology: Fundamentals and Languages ◾ 59
2.11 Ontology Tools
is section presents some useful Semantic Web software tools. ey may be clas-
sied into three major categories: ontology and metadata editors, plug-ins, and
inference tools. We will also take a short look at useful reasoning tools.
2.11.1 Protégé
Protégéis a free, open sourceontologyeditor and knowledge acquisitionsystem.
It was developed atStanford University in collaboration with theUniversity of
Manchester in 1987. It resembles Eclipse and acts as a basis for which a range
of other projects suggests plug-ins. e application is written inJavaand heavily
usesSwingto create a complex user interface.
e objective of Protégé I was to assist the knowledge acquisition process by
helping engineers build tools that would help experts in knowledge acquisition.
Protégé II is an extension of the project supporting ontology, problem-solving
mechanisms, and mapping among ontologies. It is the most popular Semantic
Web ontology editor. Its extensible open-source platform supports several ontol-
ogy formats including CLIPS (Protégé’s native format), various XML dialects,
databases, DAML+OIL, and RDFS. Storage plug-ins for Unied Modeling
Language (UML) and OWL were added recently. Protégé supports two ways
of modeling ontology—via Protégé–Frames and Protégé–OWL editors. e
Protégé–OWL plug-in extends Protégé to a graphical software development
environment that supports writing of OWL ontology. e Protégé Website pro-
vides more than 50 plug-ins. Protégé–OWL Viz is a major tool developed by the
CO-ODE group to allow visualizing and navigating through the class hierarchy
of an OWL ontology.
2.11.2 OntoEdit
OntoEdit is an ontology editor developed by the On-To-Knowledge project. It
implements an ontology construction process of three steps: requirement specica-
tion, renement, and evaluation. e editor stores the ontology conceptual model
in the Sesame repository and produces RDF concrete representations.
2.11.3 KAON2
KAON2 is a reasoner and ontology management API for OWL DL and (parts of)
SWRL. Motik wrote it in Java in cooperation with the University of Manchester,
FZI, and the University of Karlsruhe. Its features include an API for programmatic
management of OWL-DL, SWRL, and F-Logic ontology; a stand-alone server
providing access to ontology in a distributed manner using RMI; an inference
engine for answering conjunctive queries; aninterface that allows access from tools
60 ◾ Rajendra Akerkar
such as Protégé; and a module for extracting ontology instances from relational
databases. KAON2 is based onOWL-DL and F-Logic. It does not implement the
tableaux calculus. Its reasoning is implemented by dierent algorithms that reduce
a SHIQ(D) knowledge base to a disjunctive datalog program.
2.11.4 Pellet
Pellet is an open-source Java based OWL DL reasoner that may be used in conjunc-
tion with Jena and OWL API libraries. It provides a DIG interface and is an OWL
DL reasoner based on tableaux algorithms developed for expressive DLs. Itsup-
ports the full expressivity of OWL DL. As a result, the owl:oneOf and owl:hasValue
constructs can be used freely.
e basis of Pellet is the tableaux reasoner that checks the consistency of a KB,
i.e., an ABox and TBox pair. e reasoner is coupled with a datatype oracle that can
check the consistency of conjunctions of simple XML schema datatypes. e OWL
ontologies are loaded to the reasoner after species validation and ontology repair to
guarantee that all the resources have appropriate type triples, and missing type dec-
larations are added via heuristics. During the loading phase, axioms about classes
(subclass, equivalent class, disjointness axioms) are put into the TBox component
and assertions about individuals (type and property assertions) are stored in the
ABox. TBox axioms undergo the standard preprocessing of DL reasoners before
they are supplied to the tableaux reasoner.
2.11.5 FaCT++
FaCT++ is a well knownFaCT OWL-DL reasoner. It uses conventional FaCT
algorithms with special internal architectures. Moreover, FaCT++ is implemented
using C++ to create an ecient software tool and maximize portability.
2.11.6 TopBraid Composer
TopBraid is a commercial visual modeling environment for developing and man-
aging domain models and ontologies in RDF schema and OWL standards of the
Semantic Web. It has a commercially available triple store to build a multiuser
Web-accessible system that supports collaborative authoring. TopBraid is based on
the Eclipse platform and uses Jena as its underlying API to support the rapid devel-
opment of semantic applications on a single platform. It may be used to edit RDFS
and OWL les in various formats and provides scalable database backends (Jena,
AllegroGraph, Oracle 10g, and Sesame) along with multiuser support. It is a very
exible platform that enables Java programmers to add customized extensions and
develop stand-alone Semantic Web applications.
..................Content has been hidden....................
You can't read the all page of ebook, please click here login for view all page.