BLEU - Error N-gram overlaps of lower order

Question

I ran the code below

a = ['dog', 'in', 'plants', 'crouches', 'to', 'look', 'at', 'camera']
b = ['a', 'brown', 'dog', 'in', 'the', 'grass', ' ', ' ']
from nltk.translate.bleu_score import corpus_bleu
bleu1 = corpus_bleu(a, b, weights=(1.0, 0, 0, 0))
print(bleu1)

This is the error

The hypothesis contains 0 counts of 3-gram overlaps. Therefore the BLEU score evaluates to 0, independently of how many N-gram overlaps of lower order it contains. Consider using lower n-gram order or use SmoothingFunction() warnings.warn(_msg)

Can someone tell me what is the problem here? I can not find the solution on google. Thank you.

Best, DD

score 1 · Answer 1 · answered Jun 12 '20 at 21:30

I found the solution. Basically, I need a list inside a list for list 'a'. So code below will work without error.

a = [['dog', 'in', 'plants', 'crouches', 'to', 'look', 'at', 'camera']]
b = ['a', 'brown', 'dog', 'in', 'the', 'grass', ' ', ' ']
from nltk.translate.bleu_score import corpus_bleu
bleu1 = corpus_bleu(a, b, weights=(1.0, 0, 0, 0))
print(bleu1)

BLEU - Error N-gram overlaps of lower order

1 Answers1