Your index grew fast and recall fell after ingesting noisy docs. What would you clean up first?

Instruction: Explain how you would respond when content volume rises and retrieval quality drops.

Context: Tests how the candidate diagnoses the problem, chooses the safest next step, and reasons through recovery. Explain how you would respond when content volume rises and retrieval quality drops.

Official answer available

Preview the opening of the answer, then unlock the full walkthrough.

I would clean the corpus before I touched the model. Rapid index growth with falling recall is a classic sign that the retriever is drowning in low-quality or redundant material. The candidate set gets noisier, the good documents are harder to surface, and every downstream stage has a worse job....

Upgrade to view official answer

Related Questions