MediumNLPnlp-067

Top-k Embedding Cosine Ranking

Problem

Score every word in vectors by cosine similarity with query, sort by descending similarity with alphabetical tie-breaking, and return the top k as [word, similarity] pairs.

Return an empty list when k <= 0 or vectors is empty.

Hint

A zero query has no direction, so every similarity is defined as 0.0.

Test cases

case 1

in: [[1,0],{"a":[1,0],"b":[0,1],"c":[1,1],"d":[-1,0]},2]

expected: [["a",1],["c",0.7071067811865475]]

case 2

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

expected: []

case 3

in: [[1,0],{"a":[1,0],"b":[0,1],"c":[1,1],"d":[-1,0]},10]

expected: [["a",1],["c",0.7071067811865475],["b",0],["d",-1]]

case 4

in: [[0,0],{"a":[1,0],"b":[0,1],"c":[1,1],"d":[-1,0]},4]

expected: [["a",0],["b",0],["c",0],["d",0]]

case 5

in: [[1,0],{},2]

expected: []

Starter codePython

import math
def embedding_cosine_ranking_topk(query, vectors, k):
    # Your code here
    pass

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