In complex data warehouse applications, high dimensional data cubes can become very big. The quotient cube is attractive in that it not only summarizes the original cube but also it keeps the roll-up and drill-down semantics between cube cells. In this paper we study the problem of semantic summarization of iceberg cubes, which comprises only cells that satisfy given aggregation constraints. We propose a novel technique for identifying groups of cells based on bounding aggregates and an efficient algorithm for computing iceberg quotient cubes for monotone functions. Our experiments show that iceberg quotient cubes can reduce data cube sizes and our iceberg quotient cubing algorithm can be over 10-fold more efficient than the current approach.
展开▼