Chunking
TL;DR
Breaking large documents into smaller, meaningful sections.
What does this mean?
Chunking breaks long texts into smaller pieces (chunks) so they can be stored as embeddings and retrieved effectively. Chunk size directly affects search quality.
How it works
A 100-page manual is split into paragraphs or thematic blocks of 200–500 tokens each. Each chunk is stored as its own embedding.
Example
A customer service handbook is broken into individual FAQ entries so the system can find exactly the relevant section when a question is asked.
Why it matters
Good chunking determines the quality of RAG systems. Chunks that are too large dilute the search; chunks that are too small lose context.
Related terms