Ontology Query Languages for Ontology-Based Databases - Data Warehousing Design and Advanced Engineering Applications - page 237

(Seaborne and Manjunath, 2008) with the propo-
sition of SPARQL/Update. But, this language
is not yet integrated in the recommendation of
the W3C.

ables) built from a set of predefined basic path
expressions and operators (e.g., C{X} is a basic
path expression that introduces a variable X on all
instances of the class C). The WHERE clause is
used to define conditions on variables introduced
in the FROM clause. Finally, the SELECT clause
defines the variable projected in the result (like
in SPARQL).
RQL is also composed of a manipulation lan-
guage named RUL (Magiridou et al., 2005) and
a view language named RVL (Magkanaraki et
al., 2004). We present these different languages
by discussing capabilities of RQL to fulfil the
proposed requirements.

Requirement 8 (Queries
on Ontologies and on
Ontologies and Data)

Since, SPARQL considers all information as RDF
data, it can be easily used to combined ontology
and data querying. Example . Return all instances
of Resource with the classes it belongs to.

SELECT ?i ?c
WHERE {?i rdf:type sioc:Resource .
?i rdf:type ?c}

Requirement 1 (Queries at
the Ontological Level)

Explanation . The first triple of the WHERE
clause retrieves instances (?i) of the class Re-
source. The second triples retrieves classes (?c)
these instances belong to.
Notice that in this example, like in all SPARQL
queries, results depend on the triples represented
in the data source and/or the interpretation of these
triples by the query interpreter.

The data model of RQL is based on RDF-Schema.
To distinguish clearly the data, ontology and on-
tology model levels, the data model of RQL has
some restrictions compared to RDF-Schema (e.g.,
a class can not be subsumed by an ontology model
constructor). Yet, this model contains the main
constructors of RDF-Schema (class, property and
subsumption) and thus, queries at the ontological
level are possible. Example. Retrieve all instances
of the class Item RQL

The second category of Semantic Web languages
is called “RQL family”. It is mainly composed of
the languages RQL (Karvounarakis et al., 2004),
SeRQL (Broeskstra and Kampman, 2003) and
eRQL (Tolle and Wleklinski, 2004). We have
chosen to discuss the RQL language which is the
most complete.
RQL has been designed following a functional
approach similar to the object-oriented language
OQL (Cattell, 1993). Thus, simple queries consist
of function calls (e.g, SubClassOf(Resource) to
retrieve all subclasses of Resource). More elabo-
rate queries can be defined using a traditional
SELECT-FROM-WHERE syntax. The FROM
clause introduces path expressions (with vari-

SELECT I
FROM sioc:Item{I}

Explanation . The FROM clause introduces
the variable I on all (direct and indirect) instances
of the class Item. The SELECT clause projects
URI of these instances. To retrieve only direct
instances, the class Item must be prefixed with
^ (i.e., tem)

Requirement 2 (Definition of
Non Canonical Concepts)

The view language associated to RQL (RVL)
can be used to represent non canonical concepts