Recent work implementing natural logic suggests a large class of obvious inferences can be most read...