This thesis is concerned with solving combinatorial domain decision problems using incremental regret-based preference elicitation methods for interactive optimization. It is assumed that the decision maker's preferences can be represented by a parameterised scalarization function (weighted sum, OWA and Choquet integral), but the parameters (e.g., set of weights) are not known at the beginning. The active learning of the parameters is intertwined with the solution of the problem in order to learn only that part of the information about the parameter that is useful to solve the given problem. The originality of this work lies in the use of methods based on heuristic search coupled with incremental elicitation to determine the best solution for the decision maker.
At first we propose two methods for solving multi-objective combinatorial optimisation problems with imprecise preferences, the first based on local search and the second on a genetic algorithm.
We then propose two approaches to the elicitation of a linear, submodular and super-modular set function with the construction of an optimal independent subset subject to a matroid constraint. The first approach is based on a greedy algorithm and the other on the local search. In order to demonstrate the practical effectiveness of our approaches, our algorithms are numerically tested on different problems and evaluated in terms of computation time, number of queries and empirical error.