In this sense the dimension can be regarded as structurally
homogeneous.
In several situations, the structure of the elements in a category may not be the same.
Dimensions may require mixing in a single category elements representing the same
conceptual entity, but having different structure. These dimensions, which are the
focus of this chapter, have been called non-covering (Malinowsky & Zimanyi, 2004;
Pedersen et al., 2001), structurally heterogeneous (Hurtado & Mendelzon, 2002;
Hurtado et al., 2005), or ragged in the OLAP industry.
Kimball (1996), modeling products of retailers, coined the term heterogeneous
products to refer to product elements that exhibit irregularities in their dimension
structure. As an example, he stated that virtually every financial services database
has products with different structures. A bank could easily have a dozen or more
product types such as savings account, time deposit, or credit card, and each of
them may have its own set of categories and measures. Similar situations arise in
other application domains such as retail, insurance, subscription, and voyages business.
As an example, consider the product dimension for a retail warehouse given
in Figure 2, which has product elements with different structures. Products may be
musical (e.g., CDs) or electrical products (e.g., speakers, audio systems).
Pages:
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93