This seems to be a practical solution in many situations, in particular
when heterogeneity arises as an exception due to incomplete information, or when
the structure of the dimension is simple. In other cases, it seems better to repair the
structure so it can serve to guide users in query formulation. Using these techniques
we can transform heterogeneous into homogeneous data so that the standard framework
of OLAP aggregate navigation applies. We have also explained the framework
of dimension constraints, which allow to capture the mixture of structures and to
support transformations to homogeneous structures. Besides, dimension constraints
support reasoning about aggregate navigation in heterogeneous OLAP data.
There are other forms of structural heterogeneity in OLAP models. It may be found
in real data warehousing heterogeneity resulting from the nonvalidity of descriptive
attributes attached to the categories of the dimensions. These attributes allow
to describe elements of the category. As an example, we may have the attributes
CEO, number of employees, total revenue, and so forth, to describe the elements
of a category company (Jagadish et al., 1999). There is also another form of heterogeneity
that involves the structure of fact tables, since different elements may
require different measures. For example, while checking accounts have minimum
balances, overdraft limits, and service charges, saving accounts have interest paid,
and chargeable debits (Kimball, 1996).
Pages:
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131