Improved View Selection Algorithm in Data Warehouse

Abstract

In order to minimize the query processing time, a data warehouse maintains materialized views of aggregate data derived from a fact table. However, due to the expensive computing and space costs materializing the whole relations instead of part of the relations results in much worse performance. Consequently, proper selection of appropriate views to be materialized is very important to get a precise and fast response in the data warehouse. However, this view selection problem is NP-hard problem, and there have been many research works on the selection of materialized views. In this paper we propose an improved algorithm to overcome problems of existing view selection algorithms. In the presented algorithm, we first construct the reduced tables in the data warehouse using clustering method among data mining techniques, and then we consider the combination of reduced tables as the materialized views instead of combination of the original base relations. For the justification of the suggested idea, we show the experimental results in which time as well as space costs are about 1.7 times better than the conventional approaches which considered all the tuples in a relation to materialize.