Databases Reference
In-Depth Information
the language of IMDb, optimise the rewriting and use a conventional relational
database management system (RDBMS) to find the answers. (We will return to
the IMDb example in Section 6.)
The idea of OBDA was explicitly formulated in 2008 [16,26,54], though query
answering over description logic knowledge bases has been investigated since at
least 2005. Nowadays, OBDA is often deemed to be an important ingredient of
the new generation of information systems as it (
i
) gives a high-level conceptual
view of the data, (
ii
) provides the user with a convenient vocabulary for queries,
(
iii
) allows the system to enrich incomplete data with background knowledge,
and (
iv
) supports queries to multiple and possibly heterogeneous data sources.
One can distinguish between several types of OBDA depending on the expres-
sive power of description logics (DLs).
OBDA with databases:
some DLs, such as the logics of the
DL-Lite
family
), allow a reduction of conjunctive queries over ontologies to
first-order queries over standard relational databases [11,54,4,36];
OBDA with datalog engines:
other DLs encompassing logics in the
(and
OWL 2 QL
EL
fam-
, support a datalog reduc-
tion and can be used with datalog engines [63,44,50,17];
OBDA with expressive DLs
such as
ily (
OWL 2 EL
), Horn-
SHIQ
and Horn-
SROIQ
ALC
SHIQ
require some special
techniques for answering conjunctive queries; see [27,45,49,21,18,13,33] and
references therein for details.
or
In this chapter, we give a brief and easy introduction to the theory and practice
of OBDA with relational databases, assuming that the reader has some basic
knowledge of description logic. Our plan is as follows. In Section 2, we introduce
and discuss the DLs supporting first-order rewritability of conjunctive queries.
Then, in Section 3, we show how to compute first-order and nonrecursive dat-
alog rewritings of conjunctive queries over
ontologies. The size of
rewritings is discussed in Section 4. In Section 5, we introduce the basics of the
combined approach to OBDA. Finally, in Section 6, we present the OBDA sys-
tem
OWL 2 QL
, which is available as a plugin for the Protege 4 ontology editor as
well as OWLAPI and Sesame libraries and a SPARQL end-point.
Ontop
2 Description Logics for OBDA with Databases
The key notion of OBDA with databases is query rewriting. The user formu-
lates a query
)is
sometimes called an
ontology-mediated query
.) The task of an OBDA system is
to 'rewrite'
q
in the vocabulary of a given ontology
T
. (Such a pair (
T
,
q
q
in the vocabulary of the data such that,
q
and
T
into a new query
for any possible data
A
(in this vocabulary), the answers to
q
over (
T
,
A
)are
q
over
precisely the same as the answers to
A
. Thus, the problem of querying
data
A
(the structure of which is not known to the user) in terms of the ontology
T
directly. As
witnessed by the 40 years history of relational databases, RDBMSs are usually
(accessible to the user) is reduced to the problem of querying
A
