Text Summarization Actor

This Apify actor automatically generates concise summaries from longer text documents. It extracts key sentences to produce a condensed version while preserving the original meaning and essential information.

Key Features

• Automatic Processing: Generates summaries without requiring manual sentence selection.
• Quality Output: Utilizes a model trained on over 13,000 human-written summaries for accuracy.
• Content Integrity: Aims to retain the core information and intent of the source text.
• Configurable Length: Allows you to specify the desired number of sentences in the summary.

How to Use

Send a document to the actor's API endpoint. It processes the text and returns a coherent summary, making it useful for digesting lengthy reports, articles, or other documents.

Input

The actor accepts input as a JSON object containing the text to summarize and an optional parameter for output length.

{
  "text": "Your full document text goes here...",
  "output_sentences": 3
}

• text (string, required): The full text content you want summarized.
• output_sentences (integer, optional): The number of sentences for the summary. If not provided, the actor will determine an appropriate length.

Output

The results are saved to the actor's dataset in your Apify storage. You can download the data in multiple formats: JSON, JSONL, Excel, HTML, CSV, or XML.

A typical output item in the dataset is structured as follows:

{
  "summary": "The generated summary text composed of key sentences.",
  "language": "en",
  "sentenceLength": 3,
  "sentenceRanked": [
    ["2", "The second selected sentence."],
    ["0", "The first selected sentence."],
    ["4", "The third selected sentence."]
  ]
}

• summary: The final summary string.
• language: The detected language code of the input text (e.g., 'en' for English).
• sentenceLength: The number of sentences in the summary.
• sentenceRanked: An array showing the sentences used to build the summary, ranked by relevance. Each item contains the original sentence index and the sentence text.