Think about it! Improving defeasible reasoning by first modeling the question scenario