Home › Glossary › Evaluation › Citation recall

Advanced · Evaluation

Citation recall

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

TL;DR. A metric assessing what proportion of the source documents an LLM used to generate a response are actually cited within that response.

Technical Definition

A metric assessing what proportion of the source documents an LLM used to generate a response are actually cited within that response.

How it works

Citation recall measures how thoroughly an LLM acknowledges its sources. It calculates the percentage of all the documents the model consulted during generation that are actually included as citations in the final output. A high citation recall means the model is good at attributing the information it uses, whereas a low score might indicate it's omitting some sources.

Related Concepts

  • Citation precision — A metric measuring the proportion of citations in an LLM's response that accurately support the claims made within that response.

Further Reading

  • Google ML Glossary