Simple idea: if V Ì U we sample uniformly in U with a sample of size t. If s samples lie within V, then we estimate v := |V|as v = s |U|/t ,or, equivalently, by defining p :=v/|U| we estimate p as p = s/t.