Chunking
25%
Lesson 2 of 8

Chunking

Chunking splits source material into units that are small enough to retrieve but large enough to preserve meaning.

Active reading
Too large

Retrieval becomes noisy.

Too small

Meaning gets fragmented.

Useful unit

A coherent section or idea.

Chunk by meaning, not only by character count

Fixed sizes are easy to implement, but document structure often gives better boundaries: headings, paragraphs, FAQ entries or table rows can be natural chunks.

The right size depends on the question types you expect and how much context the answer needs.

Example
A refund-policy FAQ entry is often better retrieved as one complete Q&A pair than as three unrelated fragments.
Quick check

Explain this lesson back in one sentence before continuing. If you cannot, reread the example and key takeaway.

Common mistake

Do not optimize chunk size without testing real questions.

Key takeaway

A good chunk is a retrievable unit of meaning, not merely a fixed number of characters.

Next lesson

Embeddings

Continue to the next lesson.

Continue →