Multiway Pruning for Efficient Iceberg Cubing

Effective pruning is essential for efficient iceberg cube computation. Previous studies have focused on exclusive pruning: regions of a search space that do not satisfy some condition are excluded from computation. In this paper we propose inclusive and anti-pruning. With inclusive pruning, necessary conditions that solutions must satisfy are identified and regions that can not be reached by such conditions are pruned from computation. With anti-pruning, regions of solutions are identified and pruning is not applied. We propose the multiway pruning strategy combining exclusive, inclusive and anti-pruning with bounding aggregate functions in iceberg cube computation. Preliminary experiments demonstrate that the multiway-pruning strategy improves the efficiency of iceberg cubing algorithms with only exclusive pruning.