Preference Learning along Multiple Criteria