Principle:Infiniflow Ragflow Chunking Method Configuration
| Knowledge Sources | |
|---|---|
| Domains | RAG, NLP, Document_Processing |
| Last Updated | 2026-02-12 06:00 GMT |
Overview
A configuration pattern that selects the document parsing strategy for a knowledge base, determining how documents are segmented into retrievable chunks.
Description
Chunking Method Configuration allows users to select from multiple parser types that optimize document segmentation for different content formats. RAGFlow supports 14+ parser types: naive (general), paper (academic), book, laws (legal), presentation (slides), table (spreadsheets), qa (FAQ), picture (images), one (whole document), audio, email, tag, resume, and knowledge_graph. Each parser type implements specialized chunking logic that respects the structure and semantics of its target format.
Usage
Configure after creating a knowledge base and before uploading or processing documents. The parser type affects how all documents in the KB are segmented. Choose based on the dominant document format in the collection.
Theoretical Basis
Different document types have fundamentally different structures:
- Academic papers have sections, abstracts, references that should be preserved
- Legal documents have numbered articles and clauses with hierarchical structure
- Spreadsheets have rows and columns that should be kept as coherent units
- General text benefits from token-count-based splitting with overlap
The chunking method determines the parser module dispatched during document processing via the FACTORY pattern mapping.