Our model of ranking contains a ranking function R(Q, D), which compares each document with the query and computes a score. Those scores are then used to determine the final ranked list. An alternate ranking model might contain a different kind of ranking function, f(A, B, Q), where A and B are two different documents in the collection and Q is the query. When A should be ranked higher than B, f(A, B, Q) evaluates to 1. When A should be ranked below B, f(A, B, Q) evaluates to 1.

If you have a ranking function R(Q, D), show how you can use it in a system that requires one of forms f(A, B, Q). Why can you not go the other way (use f(A, B, Q) in a system that requires R(Q, D))?