Conceptual Data Warehouse Design - Data Warehouse Systems: Design and Implementation - page 91

member, denoted all . The decision of including this level in multidimensional
schemas is left to the designer. In the remainder, we do not show the All
level in the hierarchies (except when we consider it necessary for clarity of
presentation), since we consider that it is meaningless in conceptual schemas.
The identifier attributes of a parent level define how child members are
grouped. For example, in Fig. 4.2 , CategoryID in the Category level is an
identifier attribute; it is used for grouping different product members during
the roll-up operation from the Product to the Category levels. However, in
the case of many-to-many parent-child relationships, it is also needed to
determine how to distribute the measures from a child to its parent members.
For this, a distributing attribute (Fig. 4.1 g) may be used, if needed. For
example, in Fig. 4.2 , the relationship between Employee and City is many-
to-many, that is, the same employee can be assigned to several cities. A
distributing attribute can be used to store the percentage of time that an
employee devotes to each city.
Finally, it is sometimes the case that two or more parent-child relationships
are exclusive . This is represented using the symbol ' ', as shown in Fig. 4.1 h.

An example is given in Fig. 4.2 , where states can be aggregated either into
regions or into countries. Thus, according to their type, states participate in
only one of the relationships departing from the State level.
The reader may have noticed that many of the concepts of the MultiDim
model are similar to those used in Chap. 3 , when we presented the multidi-
mensional model and the data cube. This suggests that the MultiDim model
stays on top of the logical level, hiding from the user the implementation
details. In other words, the model represents a conceptual data cube.
Therefore, we will call the model in Fig. 4.2 as the Northwind data cube.

⊗

4.2 Hierarchies

Hierarchies are key elements in analytical applications, since they provide
the means to represent the data under analysis at different abstraction
levels. In real-world situations, users must deal with complex hierarchies of
various kinds. Even though we can model complex hierarchies at a conceptual
level, as we will study in this section, logical models of data warehouse and
OLAP systems only provide a limited set of kinds of hierarchies. Therefore,
users are often unable to capture the essential semantics of multidimensional
applications and must limit their analysis to considering only the predefined
kinds of hierarchies provided by the tools in use. Nevertheless, a data
warehouse designer should be aware of the problems that the various kinds
of hierarchies introduce and be able to deal with them. In this section, we
discuss several kinds of hierarchies that can be represented by means of
the MultiDim model, although the classification of hierarchies that we will
provide is independent of the conceptual model used to represent them. Since