EasyNLPnlp-347

Multiset Token F1

Problem

For token multisets, the overlap is sum over tokens of min(count_a, count_b). Precision is overlap / len(a), recall is overlap / len(b), and F1 is their harmonic mean.

Given the two token lists, return F1, or 0.0 when either is empty.

Hint

Count multiplicities rather than using set intersection.

Test cases

case 1

in: [["a","a","b"],["a","b","b"]]

expected: 0.6666666666666666

case 2

in: [["x"],["y"]]

expected: 0

case 3

in: [["a"],[]]

expected: 0

Starter codePython

def multiset_token_f1(tokens_a, tokens_b):
    # Your code here
    pass

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