Schemes of inference, conflict, and preference in a computational model of argument