Cambridge, MA, MIT Press
pages , 115--132
Margin Herbrich RGraepel TLarge KObermayer
Learning to rank using 1-norm regularization and convex hull reduction