A Visual Interface for Social Information Filtering
by John O’Donovan, Brynjar Gretarsson, Svetlin Bostandjiev, Tobias Höllerer, and Barry Smyth
The idea behind recommender systems is to provide a user with a useful recommendation. Many recommender systems use item-based or user-based collaborative filtering (CF). CF models the social process of asking a friend for a recommendation. However, without the right friends the process obviously fails. In the CF literature, such failures have been termed the “grey-sheep”, “sparsity” and “early-rater” problems. In the real world it is difficult to choose who your peers will be, as modeled in most current CF systems. Imagine you want a movie recommendation and can choose a group of peers, in addition to your existing group. You might choose Stephen Spielberg, among others if you’re in the mood for a Sci-Fi flick. However, what if you told Mr Spielberg your top 10 movies, and he hated all of them. With this knowledge, would you still want his contributions?
