What is Document chunking?

Chunking splits source documents into retrieval units that can be embedded, indexed, and supplied to a model. In practice, the value of this concept comes from understanding where it belongs in the architecture, what problem it solves, and what constraints or tradeoffs it introduces.