Preference elicitation and inverse reinforcement learning