Databases Reference
In-Depth Information
9.2. Example of Enterprise Data Architecture
As we saw in the previous chapter, the business objects
and all the semantic classes are arranged in architectural
components formed by data categories, and then by business
object domains
4
. Rather than start from scratch in order to
build this data architecture, we can usefully draw
inspiration from the one proposed by the MDM Alliance
Group which we use again here. It has a level of universality
which is sufficiently high to enable it to be used as a
foundation for semantic modeling in most companies.
Figure 9.7 is a diagram of the packages in UML. The
dotted lines represent the dependency links between the
packages.
The first level of packages reveals the business object
domains: “Catalog”, “Reality”, “Portfolio”, “Service delivery”,
“Accounting”, “Human resources”, “Supervision”. Inside each
of these packages, we discover a first, non-exhaustive list of
data categories. For example, for the package “Reality”, we
have the following data categories available: “Person”,
“Party”, “Real estate”, “Tangible assets”, “Supply capability”,
and “Geography”.
We find in the data category “Geography” the business
object “Address” and all the semantic classes which are
attached to it. In this way, the address is modeled in an
independent and re-usable way. When the data analyst
needs to describe the addresses, s/he combines its semantic
model with the data category “Geography”. Using this
principle, the same representation of the address is required
for all third parties and real estate assets.
4
. “Administrative objects” also exist, similar to business objects, but
designed with data related to the organizational aspects of the system (see
Chapter 10).
