MediumNLPnlp-247

Dense Embedding Cosine Rank

Problem

Rank one document by cosine similarity to query among the vectors keyed by id.

Sort documents by descending similarity with alphabetical tie-breaking and return the 1-based rank of target, or -1 when target is absent. A zero-norm query or vector scores 0.0.

Hint

Build the full sorted order first, then locate target in it.

Test cases

case 1

in: [[1,0],{"a":[1,0],"b":[0,1],"c":[1,1]},"c"]

expected: 2

case 2

in: [[1,0],{"a":[1,0],"b":[0,1]},"z"]

expected: -1

case 3

in: [[0,0],{"a":[1,0],"b":[0,1]},"b"]

expected: 2

case 4

in: [[1,1],{"x":[1,1],"y":[1,0]},"y"]

expected: 2

case 5

in: [[1,1],{},"a"]

expected: -1

Starter codePython

import math
def dense_cosine_rank(query, vectors, target):
    # Your code here
    pass

5 test cases · run your code against them in the editor