关键词: 物化视图, 分布式视图选择, 混合蛙跳算法, 进化算法

Abstract: To solve the View Selection Problem (VSP) under distributed scenario,a Shuffled Frog Leaping Algorithm (SFLA) based solution was presented,which was used to accelerate data warehouse queries through pre-computation and reduce the maintenance cost.After formalizing the DVSP model,the reason that the basic frog leaping rule was not applicable to DVSP was analyzed,and the method that using the recombination operator of genetic algorithm (GA) as an alternative was proposed in the local search procedure.The GA mutation operator was extended to improve the search ability under resource constraints while maintaining the diversity of frog population.A heuristic repair strategy was also proposed to deal with the infeasible solutions generated during the evolution.Experimental results showed the advantages of improved SFLA over basic SFLA and GA in terms of optimality under different constraints combinations.Compared with the penalty strategy and random ranking strategy,the heuristic repair strategy was superior in terms of solution quality and stability when the constraints were strict.

Key words: materialized view, distributed view selection, shuffled frog leaping algorithm, evolutionary algorithm