Chunking

The process of splitting documents or text into smaller segments before embedding and indexing, so each piece fits within a model's context window and can be retrieved independently. Chunk size and overlap directly affect retrieval quality in systems like Retrieval-Augmented Generation.