Question

You maintain a RAG solution over an internal engineering knowledge base in Azure AI Search. Users frequently query using exact part numbers and error codes (for example, "ERR-4821" and "P/N 9X-77"). Your current pure vector (semantic) retrieval often fails to return the document containing the exact identifier, even when it exists, because semantically similar chunks rank higher. You must improve retrieval accuracy for these exact-match queries without losing the conceptual matching that works well for natural-language questions. What should you do?

Accepted Answer

Configure hybrid search that combines keyword (BM25) and vector retrieval, and apply semantic ranking to fuse and re-rank the results.

Answer

Reduce the chunk size so each part number appears in its own chunk and lower the vector similarity threshold.

Answer

Fine-tune the embedding model on the engineering corpus so identifiers embed closer to their surrounding context.

Answer

Increase the number of nearest-neighbor results (top-k) returned by the vector query and pass all of them to the model.