PRODUCTION SCENARIO
A bank's RAG index holds 900 quarterly filings split at section headings into 600-token chunks. One chunk reads 'Net interest margin fell 20 basis points' with no issuer or quarter, so questions about a named bank's margin pull the wrong filing.
Which ingestion change targets the missing-context failure?
Answering here is anonymous. Nothing is saved unless you sign in.
Show answer and explanation
Answer: Prepend a model-written, chunk-specific context drawn from the whole filing before indexing
Contextual retrieval has a model write a concise chunk-specific context of roughly 50 to 100 tokens from the whole document and prepends it to the chunk before embedding and BM25 indexing, which cut the top-20 retrieval failure rate by 49 percent in Anthropic's tests. Bigger chunks, a keyword index, or a stronger embedding model all still work from chunk text that never names the issuer or the quarter.
Official Source: https://www.anthropic.com/engineering/contextual-retrieval ↗