Maybe the fundamental reason why exactly the same approach can't be used for
item- and user-based CF is that the tasks are not precisely symmetrical.  If
they were, and if in user-based CF we recommend the best items, then in
item-based CF we would be recommending the best *users*!
-- 
View this message in context: 
http://lucene.472066.n3.nabble.com/GenericUserBasedRecommender-vs-GenericItemBasedRecommender-tp1565019p1565201.html
Sent from the Mahout User List mailing list archive at Nabble.com.

Reply via email to