Efficient Materialized View Selection for Multi-Dimensional Data Cube Models

Abstract

Decision Support Systems help managers to make intelligent decisions by throwing complex queries on large databases. The response time to queries is a very crucial factor in governing the quality of decision support systems. The response time can be greatly improved by using query optimization techniques. A powerful query optimization technique selects only some of the views and not all views for materialization. The authors in this paper present a refined greedy selection approach using forward references to give better materialized view selection. The approach works on lattice framework of data that is capable enough to show inter dependencies of data. The choice of materialized views using the proposed approach gives a better trade off in terms of space/benefits, which is proved from the experimental results. The refined greedy selection approach is independent of space constraint and depends on number of passes entered by the user. The view selection is further enhanced by including space constraints to the results of greedy and refined greedy approach using knapsack implementation.