SVD Recommendations using Tensorflow

Recommendation system based on the user-item matrix factorization have become more and more important thanks to powerful and distributable algorithms like ALS, but sometimes the number of users and/or items is not so huge and the computation can be done using directly a SVD (Singular Value Decomposition) algorithm. In this post, we’re going to discuss a very simple implementation based on Tensorflow. The first element to define is the user-item matrix. Normally this matrix associates each user with each product through a rating (0 means that no rating has been provided) or explicit feedback. This approach is effective whenever it’s possible to ask the user to rate the items, but there are many situations where this isn’t possible. In all those cases, an implicit feedback can be employed. For example: For a movie/video, it’s possible to consider how many times the user watched it and the average duration For a […]