Incremental collaborative filtering for highly-scalable recommendation algorithms

摘要: Most recommendation systems employ variations of Collaborative Filtering (CF) for formulating suggestions items relevant to users' interests. However, CF requires expensive computations that grow polynomially with the number users and in database. Methods proposed handling this scalability problem speeding up formulation are based on approximation mechanisms and, even when performance improves, they most time result accuracy degradation. We propose a method addressing incremental updates user-to-user similarities. Our Incremental (ICF) algorithm (i) is not any gives potential high-quality (ii) provides recommendations orders magnitude faster than classic thus, suitable online application.