Databases Reference
In-Depth Information
be presented in an international manner, to allow us to have
a 360° view of the modeling approach.
For pedagogical purposes, we are firstly going to examine
a data model which does not respect the objectives of
semantic modeling, although this model is based on UML
notation and is suggested by a standards setting body.
9.1.1.
Non-compliant version of semantic modeling
Figure 9.1 presents the address data model suggested by
OASIS, a well-known standards setting body
2
.
We notice many functional dependency links of the data
straight away. For example, there are relationships which
are not represented between department (“Department”), the
name of the town (“CityName”) or the postal zone
(“PostalZone”). As this information is collected directly in the
address class (“Address”), it is impossible to formally express
the referential integrity constraints which exist between
them. The data model is therefore impoverished; it is not a
semantic data model for it lacks an element which relates to
the rules of integrity.
This functional dependency link leads to the maintenance
of duplication of data since, for example, the name of the
town (“CityName”) is duplicated as many times as the
address. If the name of the town needed to be modified, we
cannot see how the updating of the addresses in this town
could be processed efficiently.
Unfortunately, our first conclusions about this modeling
will not always be identified as problems in the application it
is used for. It is likely that the objective of this modeling is to
rationalize data flows between systems. We therefore have a
2
. Address model extract from OASIS'
Universal Business Language 1.0;
see http://docs.oasis-open.org/ubl/cd-UBL-1.0/
