Prioritized data cube selection for view materialization in data warehouse

Abstract

A data warehouse is a massive data repository, cleaned and transformed from disparate newlinedata sources to enable strategic decision-making support. Decision-makers issue OLAP newlinequeries, which are complex and ad hoc, requiring aggregated results and expecting a newlinequick response from the data warehouse. Typically, answering such queries require newlinehuge amounts of data processing resulting in high response time. Data warehouse newlinepresents a solution to speed up the query processing time by storing pre-computed newlineaggregated results in the form of materialized views. Due to storage limitations, it is newlineimpossible to store all the possible views. This raises the need to find a set of views that newlineminimizes query processing cost constrained under space limitations and is popularly newlineknown as the view selection problem [Hari96, Wido95]. newlineVarious frameworks, such as MVPP [Zhan01], AND-OR DAG [Gupt05], and lattice newlineframework [Hari96], have been designed in the literature to represent the search space newlineconsisting of the possible views. Lattice framework consisting of data cubes proves to newlinebe most suitable for the data warehouse environment. Hence, it is more relevant to name newlinethe view selection problem as the cube selection problem. View selection being an NP newline newline newline newline newline newline newline newline