Designing Data Marts from XML and Relational Data Sources - Data Warehousing Design and Advanced Engineering Applications - page 71

Figure 8. Integrated relational schema issued from e-Ticket and hotel room booking

a relationship is not the concatenation of all its
foreign keys, that is, this primary key can be an
artificial attribute such as a sequential number; or
2) when the primary key of a relationship is the
concatenation of attributes coming from empty
entities. Such attributes are never foreign keys
since an empty entity ( i.e. , entity reduced to its
key) never transforms into a relation. For more
details about empty entities, the reader is referred to
(Feki, & Hachaichi, 2007-b) where an illustrative
example is given. Table 1 shows the classification
of the relations presented in Figure 8.

In addition, they keep track of the origin (table
name, column name, data type and length…) of
each multidimensional concept in the generated
DM schemas. This traceability is fundamental as
we intend to help automating the generation of
ETL (Extract Transform and Load) procedures to
load the designed DM.

Fact Identification

To identify facts, we exploit our previous table
classification and build a set of facts of first-
relevance level from tables representing relation-
ships and then, a set of facts of second-relevance
level issued from tables representing entities.
This distinction in analysis relevance levels is
very useful; it assists the DW designer during the
selection from several generated facts those facts
that have a higher analysis potentiality.
In fact, in all DW design approaches it has
been unanimously accepted that a business ac-

DM SCHEMA CONSTRUCTION

This third design step builds DM schemas mod-
eled as stars through the application of a set of
extraction rules defined for each multidimen-
sional concept. Our rules have the merit to be
independent of the semantics of the data source.