Database Reference
In-Depth Information
Figure 1. The onion model of domain Ontology
without redundancy. In CCOs (for exam-
ple, (IEC61360-4, 1999)), information is
represented in terms of classes and proper-
ties that have unique identifiers.
constructors to relate different conceptualizations
made on this domain. Finally, a LO may provide a
natural language representation of NCCO or CCO
concepts, possibly in the various natural languages
where these concepts are meaningful.
When they are designed according to the onion
model, ontologies have always a canonical layer,
they may have a non canonical layer and they
always have a minimum of linguistic aspects and
in particular, terms that denote the represented
concept. In this chapter, we focus on the exploita-
tion of such ontologies in databases.
Non Conceptual Canonical Ontologies
(NCCOs) contain not only primitives con-
cepts (canonical) but also defined concepts,
i.e. those concepts for which the ontology
provides a complete axiomatic definition
by means of necessary and sufficient con-
ditions expressed in terms of other - primi-
tive or defined - concepts.
Linguistic Ontologies (LOs) define terms
appearing in the universe of discourse of
a given domain. Relationships between
terms are captured in a semi-formal way
by linguistic relationships (e.g, synony-
my or hyponymy). An example of LO is
Wordnet.
Extension of the ANSI/SPARC
Architecture with Ontologies
The major objectives of a database are to ensure
an efficient management of data and to provide
access to data independently of their physical
representation. The ANSI/SPARC architecture
has been proposed to fulfil these objectives. It
distinguishes two main access levels:
These three categories of ontologies can be
combined into a layered model, called the Onion
Model, shown in Figure 1. At the heart of this
model is a CCO. It provides with a formal basis to
model and to exchange efficiently the knowledge
of a domain. From primitive concepts of the CCO,
a NCCO can be designed. This NCCO provides
the
physical level which defines how data
are stored and managed using a file man-
agement system;
the
logical level which defines how data are
Search WWH ::




Custom Search