Extract Tags

Extract Tags Processor #

Extracts structured keyword tags from a document’s content using a language model.

Documents that have not yet been summarized are skipped. Run document_summarization before this processor.

Requirements #

A configured language model provider is required. Set model_provider and model in the processor config, or configure a default language model in the application settings.

Configuration #

ParameterTypeRequiredDefaultDescription
message_fieldstringNomessagesPipeline context key for the input messages
output_queueobjectNonullQueue to push processed documents to
model_providerstringNo(app default)Language model provider ID
modelstringNo(app default)Language model name
model_context_lengthintNoModel context window size in tokens (minimum 4000)
llm_generation_langstringNo(app default)BCP 47 language tag for generated content (e.g. en-US, zh-CN)

Example #

- extract_tags:
    model_provider: openai
    model: gpt-4o-mini
    output_queue:
      name: "documents_tagged"
Edit Edit this page