Text Summarization: LLM

The Text Summarization:LLM node uses a language model to generate summaries, key points, and extracted entities from input text. This node allows you to control the summarization granularity and focus areas by specifying chunk sizes, summary lengths, key point limits, and entity extraction counts. 

Inputs

  • Text – Raw or structured text input to be summarized

Outputs

  • Documents – The full structured summary output, often in JSON or object format
  • Text – A plain-text version of the generated summary

Configuration Fields

  • Number of chunks to summarize after the document is split – Defines how many parts the document should be split into before summarization
    • Example – 5
  • Number of words in each summary – Controls the length of the summary generated for each chunk
    • Set to 0 to disable summarization.
    • Example – 100
  • Number of words in each key point – Sets the word limit for each extracted key point
    • Set to 0 to disable key point generation.
    • Example – 25
  • Number of entities to extract from the document – Limits the number of named entities (example people, locations, organizations) to extract
    • Set to 0 to disable entity extraction
    • Example – 10

Steps to Configuration

  1. Drag the Text – Summarization node into the workspace
  2. Connect a source node to the Text input
  3. Fill in each configuration field depending on the desired summarization strategy
  4. Click SAVE to apply the configuration

Example

    • Field
    • Chunks
    • Summary Words
    • Key Point Words
    • Entities to Extract
    • Value
    • 3
    • 150
    • 20
    • 5
    • This configuration will break the input into 3 chunks, generate 150-word summaries, highlight up to 5 entities, and produce 20-word key points per chunk

Notes

  • Set any numeric field to 0 to skip that type of output
  • This node is commonly followed by storage, visualization, or export nodes depending on the pipeline goal