Page 68 -

A Company and its Data 29

This repository is not the focus of this book. It is a
repository based on a traditional data model which requires
heavy software developments which are too much of a
burden for the responsiveness and transparency needs of
MDM.

1.5.1.2. Semantic data repositories
A semantic repository is richer than a transactional
repository. The semantic model surpasses the classic
relational model; it welcomes all the knowledge of the data,
especially validation rules and behavioral variations
depending on the use contexts of the data model. The
semantic model enables a greater business user input in the
knowledge specification.

To put a semantic model in place, you need an MDM tool
capable of handling it in order to automatically obtain the
data repository and all its data governance functions: input
and data consultation screens, version management, context
management, audit trails, access rights management, etc. It
is no longer necessary to develop specific software to put the
data repository in place.

The wealth of the semantic model makes the IT
optimization of performance harder to ensure than with a
classic relational data model. On the other hand, an MDM
system does not have as many optimization needs as
transactional data, for the following reasons:
– the volume/quantity of reference and master data are
less important than for transactional data. On average, they
represent 20% of a company’s data;
– MDM users are the ones concerned with reference
and master data governance. There are far fewer of them
than there are of usual IT system users;
– access to an MDM system in real time is not
systematic. In most cases, the repository flows into