Conceptual Data Warehouse Design - Data Warehouse Systems: Design and Implementation - page 98

levels in a conceptual schema reduces the number of its elements without
losing its semantics, thus improving readability. In order to unambiguously
define the levels composing the various hierarchies, the hierarchy name must
be included in the sharing level for hierarchies that continue beyond that
level. This is the case of StoreLocation and SalesOrganization indicated on
level State .

City Country

CityName
CityPopulation
CityArea
...

CountryName
Capital
CountryArea
GDPGrowth
...

Store State

StoreNumber
StoreName
StoreAddress
ManagerName
...

StateName
StatePopulation
StateArea
...

SalesDistrict SalesRegion

DistrictName
Representative
ContactInfo
...

RegionName
Responsible
RegionExtent
...

Fig. 4.11 An example of parallel dependent hierarchies

Even though both alternative and parallel hierarchies share some levels and
may be composed of several hierarchies, they represent different situations
and should be clearly distinguishable at the conceptual level. This is done
by including only one (for alternative hierarchies) or several (for parallel
hierarchies) hierarchy names, which account for various analysis criteria. In
this way, the user is aware that in the case of alternative hierarchies, it is not
meaningful to combine levels from different component hierarchies, whereas
this can be done for parallel hierarchies. For example, for the schema in
Fig. 4.11 , the user can safely issue a query “Sales figures for stores in city A
that belong to the sales district B .”
Further, in parallel dependent hierarchies, a leaf member may be related
to various different members in a shared level, which is not the case for
alternative hierarchies that share levels. For instance, consider the schema
in Fig. 4.12 , which refers to the living place and the territory assignment of
sales employees. It should be obvious that traversing the hierarchies Lives
and Territory from the Employee to the State level will lead to different
states for employees who live in one state and are assigned to another. As a
consequence of this, aggregated measure values can be reused for shared levels
in alternative hierarchies, whereas this is not the case for parallel dependent
hierarchies. For example, suppose that the amount of sales generated by
employees E1 , E2 ,and E3 are $50, $100, and $150, respectively. If all
employees live in state A , but only E1 and E2 work in this state, aggregating