| As the information age explodes with data, On-line Analytical Processing (OLAP) has attracted a great deal of attention in the research communities. At the heart of OLAP is the data cube, a relational construct that supports true multi-dimensional data analysis.; Over the past decade, a number of sequential algorithms for efficient data cube construction have been presented. However, most of the algorithms are specifically designed for data set that fits in main memory. In order to reduce the expensive external memory sorting cost, Partitioned-Cube algorithm, which works with the Memory-Cube algorithm, is proposed. In this thesis, we are interested in computing the data cube efficiently when the data set is larger than the main memory. The adapted version of Partition-Cube algorithm and Memory-Cube algorithm are presented.; As noted, most of the existing algorithms are designed for implementation on sequential machines. Given the expense of such computation, another focus of our research is to compute the data cube in parallel. We present two approaches to parallelizing the Partitioned-Cube algorithm and the Memory-Cube algorithm. |
