CCAR-P

Question

You are assessing retrieval-strategy claims used by a peer team.

For each claim, select Yes when it is generally accurate; otherwise, select No.

Yes or No

Statements	Yes	No
Dense vector retrieval is well suited to recovering paraphrased intent against unstructured text.
Sparse term-matching retrieval recovers exact terms (such as identifiers and rare tokens) more reliably than dense alone.
Structured query retrieval is the right fit for filter-and-aggregate workloads against a relational store with schema and indexes.
Hybrid retrieval is identical to dense-only retrieval and shares the same recall characteristics.
Random sampling of a data store is a sound retrieval strategy for production Q&A.

Explanation

Dense vector search captures semantic similarity, which supports paraphrase retrieval. Sparse lexical retrieval is strongest for exact strings, identifiers, and rare terms. Relational queries and indexes support deterministic filtering and aggregation. Hybrid retrieval combines sparse and dense approaches, giving it different retrieval and recall behavior from dense-only search. Random sampling provides no relevance guarantee, so it is not suitable for production question-answering retrieval.

CCAR-P

AstroTutor

Notes

Readiness

QuestionQ36

Community Discussion