What is Chunking?
Chunking is the process of splitting large documents or web pages into smaller, semantically meaningful segments. This is a critical step in building RAG-based AI systems.Why Chunking Matters
•Context windows — LLMs have token limits; smaller chunks fit better
•Precision — Smaller chunks improve retrieval accuracy
•Relevance — Only the most relevant portions are sent to the AI
Chunking Strategies
| Strategy | Description | Best For ||----------|-------------|----------|
| Fixed-size | Split every N characters/tokens | Simple documents |
| Semantic | Split at paragraph/section boundaries | Structured content |
| Recursive | Progressively split using multiple separators | General purpose |
| Sentence | Split at sentence boundaries | FAQ content |
Optimal Chunk Size
There's no one-size-fits-all answer, but common guidelines:•200-500 tokens for Q&A and support chatbots
•500-1000 tokens for detailed documentation
•100-200 tokens for FAQ-style content