Databases Reference
In-Depth Information
WebNR/SD has a number of operators for Web data integration. Integration of
the Web, structured documents, and relational databases is achieved by combining
the above operators. The converters are classified into
Unpack
and
Pack
converters.
We explain primitive Unpack and Pack operators here.
Unpack
Unpack
(U)
constructs sub-relations which store text elements originally contained
in SD values. We use expressions of
the region algebra
15)
to specify which elements
are to be contained in the sub-relations. In this context,
regions
correspond to
elements in SD values. A region algebra expression represents a set of regions. For
example, region algebra expression “name” returns the set of “name” elements.
Figure 9 gives an example of Unpack, where
It constructs sub-relations for attribute C (with sub-attributes O and D) in relation
r
2
. Attribute D includes elements which are extracted from SD values in attribute
B of
r
1
, according to the specification
name
fielders
. They represent “name”
elements which are contained in “fielders” elements.
SD values in attribute
B
of relation
r
2
contain
SD references,
denoted by “
&x
.
n;
.”
“x” is called a header. The SD references refer to SD values stored in the subrelation
C. In
r
2
, we call SD values in attribute
B masters,
and those in attribute
D derivatives
.
⊂
Pack
Pack
(P)
constructs SD values from sub-relations containing elements. Figure 9
gives an example of Pack, too, where
This Pack restores original SD values in attribute
B
from sub-relations in attribute
C
of
r
2
and masters in attribute B. The masters are used as templates, and SD
references are replaced with elements in derivatives.
With converters, WebNR/SD features the following capabilities: (1) WebNR/
SD provides a framework for
symmetrically
and
dynamically
amalgamating
relational structures and structured documents. The converters transform SD values
into relational structures and vice versa. (2) The converters enable
selective
conversion
between SD values and nested relational structures. For example, given
a collection of documents with different structures, we can extract common sub-
structures and represent them in nested relational structures, ignoring detailed
structures.
4.3
AQUA
AQUA
37)
is a visual facility for both the data integration and the data layout
specification in InfoWeaver. In other systems, the user is usually required to adopt
