Abstract
In this thesis, we consider problems at the intersection of two areas: databases and knowledge bases. We focus on two equivalent problems on conjunctive queries with negation: query containment and query answering with boolean queries while making the open-world assumption. We reformulate these problems as a deduction problem in a first-order logic fragment. Then we refine existing algorithm schemes and propose new algorithms. To study and compare them experimentally, we propose a random generator and we analyze the influence of parameters on problem instances difficulty. Finally, we analyze the improvements brought out by our refinements and we compare experimentally the algorithms.