Expressing Preferences Declaratively in Logic-based Agent Languages