Introduction
How similar are two documents, two users, two images? In ML we usually answer by measuring the angle between their vectors: a small angle means "pointing the same way" regardless of length. This topic defines angles with inner products and introduces cosine similarity, the workhorse of search and recommendation.