BLEU Score
Domain: Evaluation
🎯 Motivation
Bilingual Evaluation Understudy - measures how close machine translation is to human translation. N-gram overlap.
📋 Concrete Example
Translation: ‘The cat sits’ vs reference ‘The cat is sitting.’ BLEU measures n-gram overlap. Higher BLEU = closer to human translation.
🔗 Analogy
A teacher comparing your translation to the answer key - how many words match? How many phrases match? BLEU scores translation quality similarly.