Conceptual Data Warehouse Design - Data Warehouse Systems: Design and Implementation - page 94

a situation is usually captured in an ER model using the generalization
relationship studied in Chap. 2 . Further, suppose that measures pertaining
to customers must be aggregated differently according to the customer type,
where for companies the aggregation path is Customer

→ Sector → Branch ,

while for persons it is Customer Branch . To represent such

kinds of hierarchies, the MultiDim model has the graphical notation shown in
Fig. 4.6 a, where the common and specific hierarchy levels and also the parent-
child relationships between them are clearly represented. Such hierarchies are
called generalized hierarchies .

→ Profession →

a

Sector

SectorName
Description
...

Customer

Branch
BranchName
Description
...

CustomerId
CustomerName
Address
...

Profession
ProfessionName
Description
...

b

Branch branch 1

Sector/
Profession

sector A sector B profession A profession B

.. . . ..

Customer company Z company K person X person Y

Fig. 4.6 A generalized hierarchy. ( a )Schema.( b ) Examples of instances

At the schema level, a generalized hierarchy contains multiple exclusive
paths sharing at least the leaf level; they may also share some other levels, as
showninFig. 4.6 a. This figure shows the two aggregation paths described
above, one for each type of customer, where both belong to the same
hierarchy. At the instance level, each member of the hierarchy belongs to
only one path, as can be seen in Fig. 4.6 b. We use the symbol '

⊗

' to indicate
that the paths are exclusive for every member. Such a notation is equivalent
to the xor annotation used in UML. The levels at which the alternative paths
split and join are called, respectively, the splitting and joining levels .
The distinction between splitting and joining levels in generalized hier-
archies is important to ensure correct measure aggregation during roll-up
operations, a property called summarizability, which we discussed in Chap. 3 .