BLEU (Bilingual Evaluation Understudy) was proposed by Kishore Papineni et al. in 2002. The score is calculated by taking the geometric mean of modified n-gram precision and multiplying it by a length penalty. The score ranges from 0 to 100 and is typically reported as a percentage or a score between 0 and 1, where higher values are better. It is a classic automatic metric for machine translation and text generation, but it is insensitive to synonymous expressions and is often used in conjunction with semantic metrics.