Flash Cards · Retrieval

Flash Card: Hierarchical Chunking in One Line

July 19, 2026 · 3 min read

Exam-style

A Bedrock Knowledge Base ingests 600-page equipment manuals with numbered procedures. Small fixed chunks retrieve the right step but the answer stops halfway through the procedure; large chunks carry the whole procedure but stop being retrieved at all. Which chunking strategy resolves this?

Reveal the answer

D. Hierarchical chunking, matching on the small child chunk and returning its parent chunk to the model

Hierarchical chunking indexes small children for precise matching but hands the model the larger parent that contains them, so the embedding stays focused while the context stays whole. Every other option picks a side of the size trade rather than resolving it: bigger fixed chunks and semantic chunking both dilute the embedding again, and a wider top-k only helps if the neighbouring chunks happen to score well enough to come along.

Generative AI Development · part of The Exam Room

Q. Answers get cut across chunk boundaries in a long structured PDF. Best Bedrock KB chunking?

A. Hierarchical chunking: match the small child chunk for precision, then return the larger parent chunk for context. It resolves the size trade instead of picking a side of it.

Why? Small chunks match precisely but lose context; large chunks dilute the embedding. Hierarchical gets both.

These posts are LLM-aided. Backbone, original writing, and structure by Craig. Research and editing by Craig + LLM. Proof-reading by Craig.