Chunking
Chunking is the process of splitting source documents into smaller retrieval units before embedding them. The chunk size and boundary strategy determine how precisely a retriever can locate a relevant fact, balancing recall, precision, and embedding cost across a knowledge base.
Synonyms: text chunking, document segmentation, passage splitting, chunk strategy
Chunking is where retrieval quality is quietly won or lost. The strategy can be a fixed token window, an overlapping sliding window, or boundaries that follow semantic structure such as headings and sections. Each chunk is embedded and indexed with metadata — source, language, timestamps, content hash — so retrieval can filter, deduplicate, and refresh incrementally. Because every downstream answer is only as good as the passage it retrieves, deliberate chunking is a prerequisite for grounded, citable responses.