6 ms·Factorizing a 28 billion elements matrix with Apache Flink10 points by jtagx 12y agosewen 12y agoUse case is computing recommendations at large scale (collaborative filtering, latent factor models)