An evolutionary approach to materialized views selection in a datawarehouse environment

摘要

A data warehouse (DW) contains multiple views accessed by queries.nOne of the most important decisions in designing a DW is selecting viewsnto materialize for the purpose of efficiently supporting decisionnmaking. The search space for possible materialized views isnexponentially large. Therefore heuristics have been used to search for annear optimal solution. In this paper, we explore the use of annevolutionary algorithm for materialized view selection based on multiplenglobal processing plans for queries. We apply a hybrid evolutionarynalgorithm to solve three related problems. The first is to optimizenqueries. The second is to choose the best global processing plan fromnmultiple global processing plans. The third is to select materializednviews from a given global processing plan. Our experiment shows that thenhybrid evolutionary algorithm delivers better performance than eithernthe evolutionary algorithm or heuristics used alone in terms of thenminimal query and maintenance cost and the evaluation cost to obtain thenminimal cost