context recall
An eval metric for RAG retrieval: the share of the facts a reference answer needs that appear in the retrieved context.
In more detail
Low recall means the answer could not be complete whatever the generator did, so fix retrieval before the prompt. Ragas computes it by checking whether each statement of the reference answer can be attributed to the retrieved chunks.
Related terms
- context precision — An eval metric for RAG retrieval: whether the relevant chunks were ranked above the irrelevant ones in what was retrieved.
- recall at k — The share of the relevant passages for a query that appear in the top k results.
- faithfulness — An eval metric for RAG: the share of claims in an answer that the retrieved context supports.
Taught in
- Evaluating RAG And Agents — AI Engineering: LLM Apps, RAG & Agents