X-WACoDa - Data Warehousing Design and Advanced Engineering Applications - page 47

Figure 3. DW-model.xml graph structure

straints 1 . The dw-model.xml document defines
the multidimensional structure of the warehouse.
Its root node, DW-model , is composed of two
types of nodes: dimension and FactDoc nodes.
A dimension node defines one dimension, its
possible hierarchical levels ( Level elements) and
attributes (including their types), as well as the
path to the corresponding dimension d . xml docu-
ment. A FactDoc element defines a fact, i.e., its
measures, virtual key references to the correspond-
ing dimensions, and the path to the corresponding
facts f . xml document.
Figure 4(a) represents the facts f . xml documents'
graph structure. Its equivalent XML Schema is
available on-line 2 . A facts f . xml document stores
facts. The document root node, FactDoc , is com-
posed of fact subelements that each instantiate a
fact, i.e., measure values and dimension virtual
key references. These identifier-based references
support the fact-to-dimension relationship.
Finally, Figure 4(b) represents the dimension d .
xml documents' graph structure. Its equivalent
XML Schema is available on-line 3 . A dimension d .
xml document helps instantiate one dimension,
including any hierarchical level. Its root node,
dimension , is composed of Level nodes. Each one
defines a hierarchy level composed of instance

nodes that each define the level's member at-
tribute values. In addition, an instance element
contains Roll-up and Drill-Down attributes
that define the hierarchical relationship within
dimension d . More precisely, these multivalued
virtual key references help link sets of instances
of a given dimension hierarchy level to their
aggregate in the next level ( Roll-up ), and vice-
versa ( Drill-Down ).

SOFTWARE PLATFORM

In order to support experimentations and projects
that validate our XML complex data warehousing
approach (X-WACoDa), we have developed a
software platform we also named X-WACoDa by
extension. We detail in this section its architecture,
as well as a case study that illustrates our whole
approach and exploits the software platform.
X-WACoDa's architecture is represented in
Figure 5. It consists of three components that are
further detailed in the following sections:

1. An ETL component that allows to extract

complex data representations in an homoge-
neous XML format and to integrate them into