It would be great if future versions of AMIE could compute the lift of the rules, i.e., P(B=>r(x, y)) / P(r(x, y)).