SLIM

SLIM is a library that implements a set of top-N recommendation methods based on sparse linear models. These models are a generalization to the traditional item-based nearest neighbor collaborative filtering approaches implemented in SUGGEST, and use the historical information to learn a sparse similarity matrix by combining an L2 and L1 regularization approach.

The SLIM library can be downloaded from here.

Literature

• SLIM: Sparse Linear Methods for Top-N Recommender Systems, Xia Ning and George Karypis, ICDM, 2011

External links

• SLIM software