Concept · Building with AI

Cosine similarity

Also called: cosine distance, cosine, vector similarity, dot product

A common way to measure how alike two vectors are by the angle between them. A score near 1 means very similar meaning, near 0 means unrelated.

For example: 'Refund policy' and 'how to get my money back' score close to 1. 'Refund policy' and 'giraffe' score much lower.

Connects to 5

Page last changed · Suggest a correction