Matrix Factorization: A technique used in collaborative filtering that is based on the idea of decomposing a matrix of user-item ratings into a low-dimensional space, it can be used to uncover latent factors that explain the observed ratings.