Materialized view selection applying differential evolution algorithm combined with ensembled constraint handling techniques

摘要

Materialized view selection problem is a NP-hard, constrained optimization problem where the pre-computation of views is censorious for query performance enhancement and expediting the data warehouse tasks. The pervasive presence of disk space and cost constraints heightens the intricacy of constrained optimization Materialized view selection (MVS) problem. Thus, the problem of MVS becomes prominent among data warehouse researchers. In the last few years, various evolutionary algorithms (EA) have been applied for the optimal selection of views. The present study handles the MVS problem using Ensembled Constraint Handling Techniques (ECHT) composed of (i) Self Adaptive Penalty (SP), (ii) - Constraint (EC) and (iii) Stochastic Ranking (SR) integratedwithDifferential Evolution (DE) algorithm. Authors have used TPC-H star schema benchmark dataset for testing. Simulated results were compared with three existing work i.e. PSO, genetic algorithm and EA and it was observed that our proposed ensemble method ECHTDEMVS, outperforms than single constraint handling methods and minimizes the total processing cost of query and is scalable.