Intermediate · NLP

ROUGE Score

Visual diagram · (in preparation) · Math · (in preparation) · Worked example · 3 difficulty levels.

TL;DR. A family of metrics that measure overlap between generated and reference text, used mainly for summarization.

Technical Definition

A family of metrics that measure overlap between generated and reference text, used mainly for summarization.

How it works

ROUGE-N measures n-gram overlap; ROUGE-L looks at the longest common subsequence. Like BLEU, it rewards lexical overlap and can miss good paraphrases, so production teams often supplement it with human evaluation or model-graded scoring.

Related Concepts

• Perplexity — A metric measuring how well a language model predicts text — lower means less 'surprised' by the data.
• BLEU Score — A metric that measures how close machine-generated text is to human reference translations.