Rag And Retrieval easy
An HR assistant indexes the last six yearly versions of the employee handbook. When asked about the current leave policy, it often quotes outdated versions because their text is nearly identical. What is the best fix?
- A Raise top-k so the current handbook's chunks are more likely to be included somewhere.
- B Instruct the LLM in the prompt to prefer whichever retrieved chunk sounds most recent.
- C Tag chunks with version metadata and filter to the current version at query time.
- D Delete all prior handbooks from the source system so only one version exists anywhere.
Answer: C. Tag chunks with version metadata and filter to the current version at query time.
Near-duplicate versions are indistinguishable to embeddings, so **metadata filtering** during retrieval is the reliable way to scope results; the cost is maintaining version fields at ingestion. Option B relies on the model guessing recency from wording, which is unreliable. Option D loses history that some users need; dropping old versions from the index is reasonable only if no one ever asks historical questions.
One system design question a day, with spaced repetition so the ones you miss come back.
Get the appRelated questions
- You must index 100M chunks using a 3,072-dimension embedding in fp32, served from RAM via HNSW. The raw vectors alone take about 1.2 TB before graph overhead. You need to cut memory several-fold while losing at most a point or two of recall@10. Which approach fits?
- A multi-tenant SaaS app stores all tenants in one vector index with a tenant_id field. Queries fetch the top 20 chunks by similarity and then drop chunks from other tenants. Small tenants often get 0-2 results. Why does this happen, and what is the fix?
- Runbooks are split into fixed 300-token chunks with no overlap. When a procedure step starts near the end of one chunk and finishes in the next, answers come back incomplete. What is the lowest-cost fix?
- Your offline eval shows recall@5 of 92% on a labeled set, so the gold passage is in the context for nearly every question. Yet human graders mark 35% of answers wrong, and the wrong answers often ignore or contradict the gold passage. Where should the team invest next?
- Your product requires every RAG answer to include clickable source citations that users can verify. How should those citations be produced?
- A runbook RAG system uses fixed 500-token chunks with no overlap. Users who ask for a multi-step recovery procedure get answers that stop at step 4 of 9, and retrieval logs show the top hit ends mid-procedure. What is the most targeted fix?