My implementation of simple algorithms for collaborative filtering.
-
Matrix factorization - try to recreate the original user-item interaction (rating) matrix by multiplying user-category and category-item matrices.
-
Autorec - try to encode a given sparse column/row of original user-item interaction matrix into a vector with non-zero values.
Necessary data-preprocessing tools are also provided. Aimed primarily at Movielens dataset.
add https://github.com/B0B36JUL-FinalProjects-2022/Projekt_litvidmi/examples directory contains detailed usage scenarios for each model.
ALSW as an optimization algorithm. Implementation of factorization machines and evaluation metrics for bayesian personalized ranking.