Aller au contenu principal

Summarization

Category: AI & ML
Node id: summarization

AI-powered text summarization with configurable type, length, and focus

What it does​

  • Abstractive and extractive summarization modes
  • Configurable output length (short, medium, long)
  • Multiple output formats (paragraph, bullet points, numbered list)
  • Focus area selection (main points, key facts, action items, conclusions)
  • Multi-provider support (OpenAI, Anthropic)
  • Summary quality metrics (word count, sentence count, compression ratio)

Use cases​

  • Summarizing clinical documents and discharge notes
  • Extracting key points from research papers
  • Generating executive summaries from meeting transcripts
  • Creating bullet-point summaries of patient histories

Runtime​

  • Performance tier: variable

Input ports​

NameTypeRequiredDescription
texttextYesText to summarize

Output ports​

NameTypeRequiredDescription
resultjsonNoMain result containing summary, language, and key_points
metadatajsonNoExecution metadata including tokens, original_length, summary_length, word_count, and sentence_count
successbooleanNoWhether summarization succeeded

Settings​

Basic​

SettingNameTypeRequiredDefaultAllowed valuesDescription
ProviderproviderenumNoopenaiopenai, anthropicLLM provider
ModelmodelstringNo""Model name (auto-selected if empty)
Summary Typesummary_typeenumNoabstractiveextractive, abstractive, bullet_pointsType of summarization
LengthlengthenumNomediumshort, medium, longTarget summary length
FormatformatenumNoparagraphparagraph, bullet_points, numbered_listOutput format
FocusfocusenumNomain_pointsmain_points, key_facts, action_items, conclusionsArea to focus on

Auth​

SettingNameTypeRequiredDefaultAllowed valuesDescription
API Keyapi_keysecretNoAPI key for the LLM provider

Advanced​

SettingNameTypeRequiredDefaultAllowed valuesDescription
TemperaturetemperaturenumberNo0.3min 0; max 2Sampling temperature
Max Tokensmax_tokensnumberNo500min 1Maximum tokens for summary

Errors​

CodeDescription
TEXT_REQUIREDNo text input was provided for summarization
SUMMARIZATION_FAILEDLLM call failed during summarization
UNSUPPORTED_PROVIDERProvider must be openai or anthropic

Runtime notes​

  • result.key_points is always an empty array — extract key points from the summary text if needed
  • Latency depends on LLM provider; treat as variable-time node when wiring timeouts
  • If model is empty, auto-selects gpt-4 (openai) or claude-3-5-sonnet (anthropic)
  • metadata.tokens shows actual API usage for cost tracking

See Build any workflow for how to put nodes together and Ports, types and data flow for how port types connect.

Quelque chose n'est pas clair ? Envoyez vos commentaires sur la documentation par courriel.