Recursive Chunking

~10 min read

Chunk first using natural separators like paragraphs or sections, then recursively split any chunk that still exceeds the size limit — preserving natural structure while still respecting a hard size cap.

Recursive Chunking is a Pro topic

Sign in, then upgrade to Pro or Power to unlock this topic and the full AI Engineering curriculum.

Key points

  • Two-level strategy: first split on natural separators (paragraphs/sections), then recursively split any chunk still over the size limit
  • A chunk that already fits the size limit after the first pass is left untouched
  • Preserves natural language flow and complete ideas, like semantic chunking, but uses structure rather than embeddings to decide initial boundaries
  • The size-limit enforcement is a second-pass safety net, not the primary boundary-setting mechanism
  • Costs more implementation and computational overhead than a single fixed-size pass, due to the multi-level check-and-split logic