Exit AIP-C01 AWS Certified Generative AI Developer - Professional
Question 4 of 5
0% complete
Q4 Single choice

A legal department is building a RAG assistant for long policy manuals. Users ask questions that require context from a parent section and exact clauses from child paragraphs. Retrieval must preserve document hierarchy and allow filtering by jurisdiction and publication date.

Which approach best meets these requirements?

  • A

    Use fixed-size chunking with small chunks and no metadata because smaller chunks always improve answer accuracy.

  • B

    Store the documents in Amazon S3 only, and pass the entire matching manual to the model for every request.

  • C

    Use Amazon Bedrock Knowledge Bases with hierarchical chunking and metadata filters for jurisdiction and publication date.

  • D

    Use Amazon Comprehend entity extraction only, and retrieve clauses by matching entity names in the user prompt.

Previous Next