Hi -

I've written a recommender algorithm in mahout that unifies recommendations
made in different stages of the recommendation process. The idea being
recommendations from run 1 are used in a new data model to generate
recommendations on different data where the recommendations from the first
run are used as user entries as opposed to item based entries.

Something like

First Run
(UserSet1,ItemSet1) outputs R1
Second Run
I take R1 to generate new recommendations as such (R1,ItemSet2)

I was curious, is there a defined or preferred method for doing this with
mahout?

Reply via email to