A Framework of Data Cube Computation in ROLAP

The computation of data cube is one important technique in On-Line Analytical Processing. Researchers have proposed many kinds of data cubes that are of different query response time and occupy varying size of space. Any more, each data cube has its own constructing algorithm. This paper analyzes the principles of normal data cube, partial data cube and condensed data cube that put their tuples into relation system, proposes to use the idea of fellowship to unify these kinds of data cube, and designs an algorithm TCUBE to obtain them. We also conduct an experiment using a real data set to verify the performance of TCUBE. The results show that TCUBE outperforms the original algorithms used to produce condensed cube.