On-Line Analytical Processing(OLAP) is an important application of database system that enables users to easily and selectively extract and view data from multidimensional views. OLAP queries are very complex with heavy use of aggregation, and hence they require enormous query response time. Materialization, which precomputes frequently asked part of a data cube, is widely used to reduce the query response time in OLAP. On materializing a part of a data cube, the selection of the data cube cells to be materliazed is critical to system performance. Existing researches assumed that the data cube is partitioned by aggregation attributes, and some of the partitions are materialized under a space constraint. However, the attribute-based partitioning doesn't represent effiectively hot regions on which the users' concerns are centralized.
In this work, we address a finer granule materialization method on data cube. The proposed method takes fractions of data cube partitioned by the dimension hierarchies as the basic unit of selection. The proposed method assorts hot regions effectively because the dimension hierarchies are designed to reflect users' analytical view in OLAP system. Our experimental results show that our method works better than attribute-based materialization method.