Chunking Strategies overview

Chunking Strategies as a cross-cutting topic — methods for splitting documents into meaningful units for embedding and retrieval. Covers fixed-size chunking with token or character windows, semantic chunking that detects topic boundaries using embedding similarity, parent-child chunking that maintains hierarchical document structure, sliding window approaches with configurable overlap, and code-aware chunking that respects AST boundaries. The choice of strategy directly impacts retrieval precision, recall, and the coherence of retrieved context passed to LLMs.