Skip to main content

Classifier Pipeline

Overview​

The Classifier Pipeline assigns one or more predefined labels to documents based on their content.

It is used to:

  • Categorize documents
  • Route documents into different workflows
  • Add semantic labels for filtering and decision-making

This pipeline is typically applied after parsing and tokenization.

What It Does​

  • Accepts text documents as input
  • Evaluates content against configured labels
  • Attaches classification results to document metadata

Classified documents can be:

  • Filtered
  • Routed to different pipelines
  • Used in workflows and agents
info

At least one label must be configured for this pipeline to function.

Using the Classifier Pipeline​

Add to DocProcessorAgent​

  • Open Pipelines
  • Select Classifier Pipeline
  • Drag and drop it into DocProcessorAgent
  • Connect it after the Parser or Tokenizer Pipeline

Configure Labels​

Define the set of labels used for classification.

Example labels:

  • Invoice
  • Receipt
  • Contract
  • Policy
  • Other

The pipeline compares each document against these labels and determines the best match.

Classifier pipeline label configuration

Output​

  • Documents include classification results in metadata
  • Each document receives one or more labels
  • Documents continue through the pipeline unchanged except for metadata

Common Use Cases​

  • Auto-categorizing uploaded documents
  • Routing documents to specific pipelines
  • Enabling label-based filtering and search
  • Supporting decision-based workflows

Output can be connected to:

  • Writer Pipeline
  • Workflow nodes
  • Conditional routing logic

Summary​

The Classifier Pipeline assigns semantic labels to documents, enabling automated routing, filtering, and processing in document-based workflows.