Frequent queries identification for constructing materialized views

A data warehouse contain a large amount of historical data to support decision making. The response time for decision making queries, which are usually long and complex, is high when processed against a large data warehouse This problem can be addressed by identifying and storing the relevant and required information as materialized views so that most future queries can be answered using them without accessing the large data warehouse. The approach presented in this paper attempts to identify such information using the queries posed in the past as they can serve as useful indicators of the queries likely to be posed in future. The approach uses queries posed in the past to create subject specific domains in a data warehouse. This is followed by identifying the frequent queries from amongst them. These frequent queries provide information that is highly likely to be accessed by future queries. Materialized views constructed using these queries would be capable of answering most future queries and thereby facilitate in decision making.