Scoring a Candidate Against a Session

A shopping site builds a session vector $s$ for the current visitor from the embeddings of the two items they interacted with. A click counts double a view:

$s = 2a + b$

where $a$ is the embedding of the item the visitor clicked and $b$ is the embedding of the item they only viewed. The system then ranks catalogue items by their cosine similarity to $s$ .

Vector	Meaning	Components
$a$	clicked item	$(1,\ 2,\ 2)$
$b$	viewed item	$(0,\ -1,\ 2)$
$c$	candidate item	$(6,\ 3,\ 2)$

What is the cosine similarity between the session vector $s$ and the candidate $c$ ?

Round your answer to two decimal places.

Scoring a Candidate Against a Session

Hints