A snowflake dimension has one table per category c, which
stores together all the rollup relations to the parent categories of c. The tables are
normalized avoiding redundant data. However, in contrast to the star dimension,
the snowflake dimension requires to process join operations over its tables, in order
to assemble tuples for browsing and aggregate query processing.
Relationship among Graph, Star, and Snowflake Dimensions
Dimensions in the graph model we presented (we will refer to them simply as dimensions,
in contrast to star and snowflake dimensions) can be easily translated to
a star dimension (Hurtado et al., 1999), provided the dimension is homogeneous
and has a single bottom category. The table has as attributes the categories of the
dimension. For each bottom element m of the graph dimension, there is a tuple in
the table composed of the ancestors of m. As an example, Table 1 shows the star
dimension corresponding to the dimension of Figure 4.
The translation of a star dimension to a graph dimension requires capturing the
rollup relations from the table. It is simply obtained by projecting the table over
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.
the categories involved in the rollup relation.
Pages:
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112