Embeddings as features, not magic (step 2/7) · unsupervised learning, embeddings, and recommenders

A policy-search tool returns a high-scoring document from the wrong department. What should the receipt do?

Embeddings as features, not magic — step 2 of 7