LSA-Based Compression of Data Cubes for Efficient Approximate Range-SUM Query Answering in OLAP

Abstract

AbstractNowadays a lot of research efforts have been done in the context of approximate query answering techniques in OLAP, which pursue the idea of compressing the data cube in order to obtain approximate answers to OLAP queries whose (approximation) error is tolerable in real-life Business Intelligence scenarios. In this chapter, we introduce a novel approximate OLAP query answering technique that is based on an innovative analytical interpretation of multidimensional data cubes, and the use of the well-known Least Squares Approximation (LSA) method in order to build the so-called analytical synopsis data structure Δ-Syn. The benefits deriving from adopting Δ-Syn within the core layer of modern OLAP server platforms is confirmed by a comprehensive experimental evaluation of the performance of Δ-Syn on both synthetic, benchmark and real-life data cubes that clearly shows the superiority of Δ-Syn in comparison with state-of-the-art approximate query answering techniques like histograms, wavelets and random sampling.KeywordsRange QueryData CubeQuery AnsweringApproximate AnswerApproximate QueryThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.