bilingual evaluation understudy

Posted 2020-10-07 rsapaper

 

BLEU is designed to approximate human judgement at a corpus level, and performs badly if used to evaluate the quality of individual sentences.

 

 

https://en.wikipedia.org/wiki/BLEU

To produce a score for the whole corpus the modified precision scores for the segments are combined using the geometric meanmultiplied by a brevity penalty to prevent very short candidates from receiving too high a score.