HierarchicalDocumentSplitter

Split documents into different block sizes, building a hierarchical tree structure.

Basic Information

• Type: deepset_cloud_custom_nodes.preprocessors.hierarchical_document_splitter.HierarchicalDocumentSplitter
• Components it can connect with:
  • Converters: HierarchicalDocumentSplitter receives documents from converters or DocumentCleaner.
  • DocumentCleaner: HierarchicalDocumentSplitter can receive cleaned documents from DocumentCleaner.
  • Embedders: HierarchicalDocumentSplitter sends split documents to document embedders.
  • DocumentWriter: HierarchicalDocumentSplitter can send documents to DocumentWriter for storage.
  • AutoMergingRetriever: Use with AutoMergingRetriever in query pipelines to merge child documents back to parent context.

Inputs

Parameter	Type	Default	Description
documents	List[Document]		List of Documents to split into hierarchical blocks.

Outputs

Parameter	Type	Default	Description
documents	List[Document]		List of hierarchical documents with parent-child relationships.

Overview

HierarchicalDocumentSplitter splits documents into blocks of different sizes, building a hierarchical tree structure. The root node of the tree is the original document, and the leaf nodes are the smallest blocks. The blocks in between are connected such that smaller blocks are children of parent (larger) blocks.

This hierarchical structure is useful for advanced retrieval techniques like auto-merging retrieval, where you retrieve small, specific chunks but can expand to larger parent chunks for more context when needed.

Key parameters:

• block_sizes: Set of block sizes to split the document into (splits in descending order)
• split_overlap: Number of overlapping units between chunks
• split_by: The unit for splitting - word, sentence, page, or passage

Usage Example

Using the Component in an Index

This example shows an indexing pipeline using hierarchical splitting with block sizes of 512, 256, and 128 words.

components:
  TextFileToDocument:
    type: haystack.components.converters.txt.TextFileToDocument
    init_parameters:
      encoding: utf-8
      store_full_path: false
  DocumentCleaner:
    type: haystack.components.preprocessors.document_cleaner.DocumentCleaner
    init_parameters:
      remove_empty_lines: true
      remove_extra_whitespaces: true
  HierarchicalDocumentSplitter:
    type: deepset_cloud_custom_nodes.preprocessors.hierarchical_document_splitter.HierarchicalDocumentSplitter
    init_parameters:
      block_sizes:
      - 512
      - 256
      - 128
      split_overlap: 0
      split_by: word
  SentenceTransformersDocumentEmbedder:
    type: haystack.components.embedders.sentence_transformers_document_embedder.SentenceTransformersDocumentEmbedder
    init_parameters:
      model: sentence-transformers/all-MiniLM-L6-v2
      batch_size: 32
      progress_bar: true
      normalize_embeddings: false
  DocumentWriter:
    type: haystack.components.writers.document_writer.DocumentWriter
    init_parameters:
      document_store:
        type: haystack_integrations.document_stores.opensearch.document_store.OpenSearchDocumentStore
        init_parameters:
          hosts:
          index: hierarchical-documents-index
          max_chunk_bytes: 104857600
          embedding_dim: 384
          return_embedding: false
          create_index: true
          similarity: cosine
      policy: NONE

connections:
- sender: TextFileToDocument.documents
  receiver: DocumentCleaner.documents
- sender: DocumentCleaner.documents
  receiver: HierarchicalDocumentSplitter.documents
- sender: HierarchicalDocumentSplitter.documents
  receiver: SentenceTransformersDocumentEmbedder.documents
- sender: SentenceTransformersDocumentEmbedder.documents
  receiver: DocumentWriter.documents

max_runs_per_component: 100

metadata: {}

inputs:
  files:
  - TextFileToDocument.sources

Parameters

Init Parameters

These are the parameters you can configure in Pipeline Builder:

Parameter	Type	Default	Description
block_sizes	Set[int]		Set of block sizes to split the document into. The blocks are split in descending order.
split_overlap	int	0	The number of overlapping units for each split.
split_by	Literal['word', 'sentence', 'page', 'passage']	word	The unit for splitting your documents.

Run Method Parameters

These are the parameters you can configure for the component's run() method. This means you can pass these parameters at query time through the API, in Playground, or when running a job. For details, see Modify Pipeline Parameters at Query Time.

Parameter	Type	Default	Description
documents	List[Document]		List of Documents to split into hierarchical blocks.