Notice that the two dimensions have the same hierarchy
domain. However, in the new dimension the products are broken up and stored
in different categories for each of the four structures of Figure 3. Figure 10 shows an
unbalanced homogeneous version for the heterogeneous dimension of Figure 4.
The model provides flexibility to define a different base fact table for each bottom
category, a single base fact table for all the categories, or both. If we consider different
fact tables, the hierarchy paths of the former are consistent. Now the cube
view CV[Department] can be correctly computed by the rollup:
ROLLUP Category TO Department
If we consider cube views defined over a single base fact table that contains all the
elements, the hierarchy paths are not consistent. In this case we need ROLLUP
operations that combine categories, as we will see in the next section.
Kimball (1996) proposes a similar solution to deal with heterogeneity, but in his
approach the homogeneous structures mixed are placed in separate dimensions.
Figure 10. An unbalanced dimension that models the Product dimension of Figure 4:
(a) hierarchy schema; (b) hierarchy domain
(a) (b)
Handling Structural Heterogeneity in OLAP 4
Copyright ?© 2007, Idea Group Inc. Copying or distributing in print or electronic forms without written permission
of Idea Group Inc. is prohibited.
Pages:
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120