Database Reference
In-Depth Information
over the materialized target
T
0
, computes
certain
M
(
Q
,
S
). This query
Q
is called a rewrit-
ing of
Q
over
T
0
;then
Q
(
T
0
)=
certain
M
(
Q
,
S
)
.
When we have such a materialized solution
T
0
and a rewriting
Q
for each query
Q
we
want to pose, we have the answer to the two key questions of data exchange. First, we
know which solution to materialize - it is
T
0
. Second, we know how to answer queries -
compute the rewriting
Q
of
Q
and apply it to
T
0
.
3.4 Bibliographic comments
Data exchange, also known as data translation, is a very old problem that arises in many
tasks where data must be transferred between independent applications (
Housel et al.
,
1977
). Examples of data exchange problems appeared in the literature over 30 years ago.
But as the need for data exchange increased over the years (
Bernstein
,
2003
), research
prototypes appeared and made their way into commercial database products. An early mo-
tivating example for much of research on data exchange was IBM's Clio system, described
by
Miller et al.
(
2001
)and
Fagin et al.
(
2009
). The theory was lagging behind until the
paper by
Fagin et al.
(
2005a
) which presented the widely accepted theoretical model of
data exchange. It developed the basis of the theory of data exchange, by identifying the
key problems and looking into materializing target instances and answering queries.
Within a year or two of the publication of the conference version of this paper, data
exchange grew into a dominant subject in the database theory literature, with many papers
published in conferences such as PODS, SIGMOD, VLDB, ICDT, etc. By now there are
several papers presenting surveys of various aspects of relational data exchange and schema
mappings (
Kolaitis
,
2005
;
Bernstein and Melnik
,
2007
;
Barcelo
,
2009
), and a short topic
dealing with both relational and XML data exchange by
Arenas et al.
(
2010b
). In addition
to the Clio system mentioned early, a new open-source system called ++Spicy has recently
been released, see the paper by
Marnette et al.
(
2011
). Much more extensive bibliographic
comments will be provided in the subsequent chapters.
The subject of data integration has also received much attention, see, for example, the
keynote by
Haas
(
2007
), the tutorial by
Lenzerini
(
2002
), and the comprehensive survey
by
Doan et al.
(
2012
). Relationships between data exchange and integration have also been
explored (
Giacomo et al.
,
2007
).
For additional background information on relational databases, complexity classes, and
automata the reader is referred to standard textbooks, for example, the topics by
Abiteboul
et al.
(
1995
);
Ullman
(
1988
);
Hopcroft and Ullman
(
1979
);
Papadimitriou
(
1994
);
Garey
and Johnson
(
1979
);
Garcia-Molina et al.
(
2001
); and
Sipser
(
1997
).
