Sean Owen wrote: > > > You mean, start with the user-rated items as candidate items, then use > a neighborhood of items around those as the basis for a similarity > computation? Yes exactly, that doesn't work. The user probably hasn't > rated much in that neighborhood. > >
Not exactly. If I understand Stanley's idea, his goal is to provide a predicted rating for a particular item, and he's asking why not calculate a neighborhood of similar items around that particular item, and then within these, find the subset of items that the user has rated and calculate the weighted average using this subset. So I don't think he's proposing neighborhoods around items that the user has rated, but around items that we want to generate a predicted rating for. But as you and I both said, this is inefficient because of the items in that neighborhood that the user won't have rated. I hope that clarification makes the rest of my post make more sense. -- View this message in context: http://lucene.472066.n3.nabble.com/GenericUserBasedRecommender-vs-GenericItemBasedRecommender-tp1565019p1566710.html Sent from the Mahout User List mailing list archive at Nabble.com.
