Boosting collaborative filtering based on statistical prediction errors

• DOI: 10.1145/1454008.1454011
• Publication Date: 2008-11-06T13:49:50Z
• Authors: Shengchao Ding; Shiwan Zhao; Quan Yuan; Xiatian Zhang; Rongyao Fu; Lawrence Bergman
• Keywords: Similarity (geometry); Boosting; Gradient boosting

User-based collaborative filtering methods typically predict a user's item ratings as weighted average of the given by similar users, where weight is proportional to user similarity. Therefore, accuracy similarity key success recommendation, both for selecting neighborhoods and computing predictions. However, computed similarities between users are somewhat inaccurate due data sparsity.