What do you get when you cross a support vector machine with matrix factorization? You get a factorization machine, and a darn fine algorithm for recommendation engines.