As explained in the approach, MiningMart uses a layer of abstraction of the business data
in order to hide sensitive details from other MiningMart users.
This layer is the conceptual level. Its name stems from the fact that on this level, the data
is described in everyday concepts rather than in terms of its technical representation.
For example, many institutions have got data about their customers.
So it could make sense to introduce the common concept ``Customer'' on the conceptual level,
where it represents the data about customers. Information about this level forms part of the
Metadata described above.
The conceptual level is the most important one for MiningMart users,
because all the data processing is described in terms of the conceptual level.
That is, whenever the customer data in the above example is accessed,
this is done via the concept ``Customer''. In contrast to this level, there is the
relational level which also forms part of the Metadata, but which contains less
abstract information about the business data. Both levels must be connected.