A Generalized Approach to Optimization of Relational Data Warehouses Using Hybrid Greedy and Genetic Algorithms

As far as we know, in the open scientific literature, there is no generalizedframework for the optimization of relational data warehouseswhich includes view and index selection and vertical view fragmentation.In this paper we are offering such a framework. We proposea formalized multidimensional model, based on relational schemas,which provides complete vertical view fragmentation and presents anapproach of the transformation of a fragmented snowflake schema toa defragmented star schema through the process of denormalization.We define the generalized system of relational data warehouses optimizationby including vertical fragmentation of the implementationschema (F), indexes (I) and view selection (S) for materialization. Weconsider Genetic Algorithm as an optimization method and introducethe technique of ”recessive bits” for handling the infeasible solutionsthat are obtained by a Genetic Algorithm. We also present two novelhybrid algorithms, i.e. they are combination of Greedy and GeneticAlgorithms.Finally, we present our experimental results and show improvementsof the performance and benefits of the generalized approach(SFI) and show that our novel algorithms significantly improve theefficiency of the optimization process for different input parameters.