A Declarative Approach to Modeling and Solving the View Selection Problem

摘要: View selection is important in many data-intensive systems e.g., commercial database and data warehousing to improve query performance. can be defined as the process of selecting a set views materialized order optimize evaluation. To support this process, different related issues have considered. Whenever source changed, built on it maintained compute up-to-date results. Besides view maintenance issue, each also requires additional storage space which must taken into account when deciding how materialize. The problem choosing materialize that speed up incoming queries constrained by an overhead and/or costs, known problem. This one most challenging problems NP-complete In distributed environment, becomes more challenging. Indeed, includes another issue decide computer nodes selected should materialized. context now additionally capacities per node, maximum global costs communications cost between network. work, we deal with centralized well setting. Our goal provide novel efficient approach these contexts. For purpose, designed solution using constraint programming for resolution powerful method modeling solving combinatorial optimization problems. originality our provides clear separation formulation modeled satisfaction easy declarative way. Then, its performed automatically solver. Furthermore, flexible extensible, easily model handle new constraints heuristic search strategies purpose. main contributions thesis are follows. First, define framework enables better understanding address thesis. We analyze state art review existing methods identifying respective potentials limits. then design context. performance experimentation results show has ability best balance computing time required findin tghe gain realized processing materializing views. will guarantee pick optimal where no limit imposed. Finally, extend latter studied under multiple resource Based extensive evaluation, outperforms genetic algorithm been