Optimisation des requêtes dans le datawerehouse

Abstract

The work presented in this memory is related to the optimization of the query in the data warehouse, more particularly the technique of the materialized views. Big problems of optimization were suggested in this technique, which deals with the problem of selection of views. The technique consists of selecting the whole of the views to be materialized to accelerate in the future the execution of the query. Constraints were required in the selection process, which is directed with the system and the user. The matérialisation of all the views accelerates the execution of any query but requires a considerable cost to maintain them, knowing that the maintenance of the views returns the results of the query more beneficial for the decision makers.Then, the problem becomes an optimization multi-objectives. This memory proposes a solution with this problem in three steps: In the first step, we used and developed a mathematical formulation.In the second step, we built a lattice of the possible views.In the third step, we applied the multi-objectives genetic algorithms on the lattice so far developed . The variante adopted in the resolution is NSGA II (Nondominated Sorting Genetic Algorithm).