I. INTRODUCTION
The method of data organization in a data warehousel4]
plays an important role for data statistics and decision analysis
and influences the efficiency of data warehouse directly. At
present, the construction of data warehouse mainly uses
methods of a dual level of granularity, partitioning datal'] and
so on, which divides the organization of data into an older level
of detail, a current level of detail, a level of lightly summarized
data and a level of highly summarized data and stores the data
fragments according to time and types. This method, which
makes it suitable for normal and general data statistics and
decision analysis, is easy to design and realize. But on the other
hand, this kind of data organization developed from the point
of view of developers ignores the needs, habits and attention of
consumers. By analyzing query condition (predicate[2])
statistics, this article is designed for the convenience of
consumers and put forwards a data organization of data
warehouse which can organize data according to the
materialized viewl3] generated by predicates of different classes,
providing a more convenient and effective way for different
consumers.
978-