Databases Reference
In-Depth Information
8.4. The drawbacks of semantic modeling
Every modeling attempt is face with three drawbacks: a
lack of return on investment, a lack of competency and the
blank page effect. This also concerns the semantic modeling
required by the MDM approach.
8.4.1.
The lack of return on investment
We have already had the opportunity to present the
conditions necessary to successfully transform models on
return on investment (see Chapter 4). The important
elements that we must bear in mind are repeated here:
-
the reference and master data models are exploited
by a Model-driven MDM system in order to automatically
deliver the data governance functions. From this stage on,
the return on investment of modeling is considerable as the
business users have a data governance tool at their disposal
directly aligned with the reference and master data models;
-
these same data models become the canonical format
of data flows in the EAI, ESB or ETL. Different to a pivot
format often too rigid and technical, this canonical format
expresses all the business knowledge of the data and can
take variations in behavior into consideration, for instance in
the data cardinality of associations
10
.The streamlining of
10
. This canonical format, in an XML schema, enriches the default
management of the data cardinalities minOccurs and maxOccurs, the values
of which are fixed and frozen in the description of data structures. Indeed,
they are themselves managed as reference data, with values that depend on
the use contexts of the canonical format. It means that the description of
data structures made in the XML Schema is contextualized depending on
the consumers. This is the same as extending the XML schema, which is
authorized by the standard, in order to add predicates, for example in xPath,
to retrieve the values minOccurs and maxOccurs depending on the use
contexts. A Model-driven MDM tool is capable of implementing this type of
agility on the data (see the MDM tool of Orchestra Networks for more
details).
