I have a question regarding Bleu score that while calculating it considers maximum number of words in machine translation output present in reference for calculating count clip divided by count of (pairs/single) occurence of words. But what if we re-arrange words in a group of four, it will still have high Bleu score, but since output is rearranged it is not proper output. How to solve it? Does the solution is already taken care by Beam search algorithm regarding ordering of words or their any other approach. Please clarify.
Hi Mohit,
Sorry for the late response. According to the paper referenced in the lecture:
“…a high-scoring candidate translation must now match the reference translations in length, in word choice, and in word order.”
Hope this helps.
Best,
Rosa