Top-Down Compression of Data Cubes in the Presence of Simultaneous Multiple Hierarchical Range Queries

Abstract

A novel top-down compression technique for data cubes is introduced and experimentally assessed in this paper. This technique considers the previously unrecognized case in which multiple Hierarchical Range Queries (HRQ), a very useful class of OLAP queries, must be evaluated against the target data cube simultaneously. This scenario makes traditional data cube compression techniques ineffective, as, contrarily to the aim of our work, these techniques take into consideration one constraint only (e.g., a given space bound). The result of our study consists in introducing an innovative multiple-objective OLAP computational paradigm, and a hierarchical multidimensional histogram, whose main benefit is meaningfully implementing an intermediate compression of the input data cube able to simultaneously accommodate an even large family of different-in-nature HRQ. A complementary contribution of our work is represented by a wide experimental evaluation of the query performance of our technique against both benchmark and real-life data cubes, also in comparison with state-of- the-art histogram-based compression techniques.