“…The Recall, Precision, and F1 scores for the reference 𝑥 and candidate 𝑥 ̂ are: Where RBERT counts the number of correctly translated words compared to the machine-translated words, PBERT counts the number of candidate translation words (unigrams) that occur in any reference translation by the total number of words in the candidate translation. The F-measure (FBERT) of the translation is equal to the multiplication of the Precision and recall divided by the addition of the Precision and Recall [21], [27][28].…”