Hi,
As far as I know, if we use Mahout to build the enterprise recommendation engine, we can precompute the itemSimilar and userSimilar by hadoop on the background(offline), then store the result inside the database. the recommendation engine can provides the recommend result online. But, I don’t think this is the best solution, in my own opinion the model-based is the best solution. But It does not currently support model-based recommenders. So ,is anybody build the model-based recommender successfully or does anybody have the better solution? Best Wishes Jack
