Learning to Rank With Bregman Divergences and Monotone Retargeting