EasyNLPnlp-307

ROUGE-2 Clipped Bigram Overlap

Problem

Return the number of clipped bigram matches between candidate and reference token lists. A bigram match consumes one occurrence in the reference, so repeated candidate bigrams cannot match more times than they appear in the reference.

Return 0 when either list has fewer than two tokens.

Hint

Count matches, not F1: each matched candidate bigram consumes one reference occurrence.

Test cases

case 1

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

expected: 2

case 2

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

expected: 1

case 3

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

expected: 0

case 4

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

expected: 1

case 5

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

expected: 0

Starter codePython

def rouge2_overlap(candidate, reference):
    # Your code here
    pass

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