Multiobjective Differential Evolution Algorithm Using Binary Encoded Data in Selecting Views for Materializing in Data Warehouse

摘要

In this paper, we define the view selection process for materializing in data warehouse as a multiobjective optimization problem. We have implemented multiobjective Differential Evolution (DE) algorithm for binary encoded data to solve this problem. In our approach, to control population in intermediate generations of the differential evolution process by maintaining diversity in solution space with necessary elitism, the solutions of intermediate generations are first ranked according to their pareto dominance levels and then the diversity among solution vectors in solution space is measured. The algorithm is found to be suitable in selecting significant representitive solutions from a large number of nondominating solutions of the view selection problem.