Explain 'Matrix Factorization' for recommendation.

Explain 'Matrix Factorization' for recommendation.

Approach: Describe decomposing a user-item matrix into lower-dimensional matrices.

A class of collaborative filtering algorithms used in recommender systems. It works by decomposing the user-item interaction matrix into the product of two lower-dimensionality rectangular matrices (user embeddings and item embeddings).

Follow-up: How does it handle missing entries in the matrix?

Key concepts: latent factors, svd, embeddings.