Designing Data Marts from XML and Relational Data Sources - Data Warehousing Design and Advanced Engineering Applications - page 75

Table 3. Dimensions for the extracted facts

Fact Dimension Identifier

D-RoomTypes RoomTypeID

F-Room D-RoomBands RoomBandID

D-Facilities RoomFaciliyID

D-Customer CustomerID

F-Payments

D-PaymentMethods PaymentMethodID

D-Customer CustomerID

D-DateBookingMade DateBookingMade

D-TimebookingMade TimebookingMade

F-Bookings

D-Room RoomID

D-BookedStartDate BookedStartDate

D-BookedEndDate BookedEndDate

D-Customer CustomerID

D-Concert CustomerID

F-Buy

D-BuyDate BuyDate

D-ConcertDate ConcertDate

structure to relational scheme; 3) Acquisition of
the relational schema ; 4) Schema integration by
applying the fusion approach presented in (Hull,
1997); 5) Conceptual class identification ; 6)
DM Conceptual design whose result can be seen
both in a tabular format (as illustrated in Figure 9
through the running example) and in a graphical
representation (Figure 10); and 7) Adaptation of
the obtained DM when the designer adjusts the
constructed DM schemas to the analytical require-

ments of decisional users. In this step, CAME is
linked to our case tool MPI-EDITOR (BenAbdal-
lah, Feki, & Ben-Abdallah, 2006) which allows the
designer to graphically manipulate the built DM
schemas and adapt them to produce well-formed
schemas (Figure 10).

CONCLUSION

In this chapter, we have presented a bottom-up/
data-driven design method for DM schemas from
two types of sources: a relational database and
XML documents compliant to a given DTD. Our
method operates in three automatic steps (Data
source pretreatment, relation classification, and
DM schema construction) followed by a manual
step for DM schema adaptation. It exploits the
recent schema/DTD version of a data source to
automatically apply a set of rules that extract all
candidate facts with their measures, dimensions
and hierarchies. In addition, being automatic, our
design method is supported by a CASE toolset
that allowed us to evaluate it through several

Table 4. Parameters for the dimensions of Table
3

Hierarchy parameters
(From finest to coarsest)

Dimension

CityID StateID CountyID

CustomerDOB

D-Customer

RoomTy peID
RoomBandID
RoomFaciliyID

D-Room

SingerID

D-Concert

CityID StateID CountyID