An Effective Drill-Down Paths Pruning Method in OLAP

Abstract

The complexity of multi-dimensional data structure affects the efficiency of OLAP, because there are too many drill-down paths to be chosen from when analysis. While most methods in the literature are associated to some specific analysis tasks, so they cannot get reasonable effect. In this paper, we proposed a new method that is irrelevant to analysis task that we try to prune the invalid drill-down operations. The vectorial angle method is employed to evaluate the validness of every drill-down operation. We give the corresponding path pruning algorithm, and it is effective that it takes the fact table as the input in only one pass scanning. The experiments show that our method is feasible, effective, sparsity-proof and skewness-proof