Collaborative Filtering and Recommender Systems Recommender Systems
The main problem solved by collaborative filtering methods/recommender systems can be phrased in a number of ways: user-based recommendations, content-based recommendation systems, and collaborative recommendation systems.
Abstract
Definitions Recommendation generation problem. Given a set of users and their (incomplete) preferences over set of items, find, for each user new items for which they would have high preferences. Typically, utility function is incomplete. Problem. The main problem solved by collaborative filtering methods/recommender systems can be phrased in a number of ways: • User-based recommendations. Given a user c find items s ′ 1 ,. .. s ′ k (such that u(c, s ′ i) is undefined), for which c is predicted to have highest utility. Recommender System: a system, which given C, S and a partial utility function u, solves one or more of the problems of recommendation generation. Content-based recommendation systems: recommend items similar to the ones preferred by the user in the past. Collaborative recommendation systems: recommend items that other users with similar preferences find to be of high utilitiy.