Incremental training in recommander

[Mathieu sgard]

Hello,

I'm working on a recommender feature in e-commerce.
Is it possible to train the mahout recommender in incremental way or the
only way is compute entire dataset when new items are added ?

Thanks