Summarization Overview

Neuredge's summarization capability helps you extract key insights and main points from long texts efficiently. Our models are optimized for accuracy while maintaining the essential context of the original content.

Features

Core Capabilities

• Flexible Length Control: Customize summary length
• Key Points Extraction: Identify main ideas and crucial information
• Multi-Language Support: Summarize content in multiple languages
• Format Preservation: Maintain important structural elements

Quota Details

• Free Tier: 100K tokens/month
• $29 Plan: 1M tokens/month
• $49 Plan: 1.5M tokens/month
• Enterprise: Custom quotas available

Getting Started

Basic Usage

import { NeuredgeClient } from '@neuredge/sdk';

const client = new NeuredgeClient({
  apiKey: 'your-api-key'
});

// Simple summarization
const summary = await client.text.summarize(
  'Your long text here...',
  { maxLength: 100 }
);

// With more options
const detailedSummary = await client.text.summarize(
  'Your long text here...',
  {
    maxLength: 200,
    format: 'bullets',
    preserveKeyPoints: true,
    language: 'en'
  }
);

Python Example

from neuredge_sdk import NeuredgeClient

client = NeuredgeClient("YOUR_API_KEY")

# Basic summarization
summary = client.text.summarize(
    "Your long text here...",
    max_length=100
)

# Advanced options
detailed_summary = client.text.summarize(
    "Your long text here...",
    max_length=200,
    format="bullets",
    preserve_key_points=True,
    language="en"
)

Advanced Features

Bullet Point Summarization

const bulletSummary = await client.text.summarize(
  'Your long text here...',
  {
    format: 'bullets',
    maxPoints: 5
  }
);

Hierarchical Summarization

const hierarchicalSummary = await client.text.summarize(
  'Your long text here...',
  {
    format: 'hierarchical',
    levels: 2,
    pointsPerLevel: 3
  }
);

Multi-Document Summarization

const documents = [
  'First document content...',
  'Second document content...',
  'Third document content...'
];

const combinedSummary = await client.text.summarizeMultiple(
  documents,
  {
    maxLength: 300,
    preserveKeyPoints: true
  }
);

Best Practices

Optimizing Results

1. Input Preparation

   • Clean and format input text
   • Remove unnecessary formatting
   • Split very long texts into logical sections

2. Length Control

   • Set appropriate maxLength
   • Use format options for better structure
   • Consider content type when choosing parameters

3. Error Handling

try {
  const summary = await client.text.summarize(
    'Your text here...',
    { maxLength: 100 }
  );
} catch (error) {
  if (error.code === 'token_quota_exceeded') {
    console.log('Token quota exceeded for this tier');
  } else if (error.code === 'invalid_request') {
    console.log('Invalid request parameters');
  }
}

Use Cases

Content Digestion

// Summarize article with key points
const articleSummary = await client.text.summarize(
  articleText,
  {
    format: 'structured',
    sections: ['overview', 'key_points', 'conclusion']
  }
);

Meeting Notes

// Summarize meeting transcript
const meetingSummary = await client.text.summarize(
  transcriptText,
  {
    format: 'bullets',
    categories: ['decisions', 'action_items', 'key_discussions']
  }
);

Research Papers

// Academic paper summarization
const paperSummary = await client.text.summarize(
  paperText,
  {
    format: 'academic',
    sections: ['abstract', 'methodology', 'findings', 'conclusion']
  }
);

Integration Examples

Workflow Automation

async function processDocument(doc: string) {
  // Generate summary
  const summary = await client.text.summarize(doc, {
    maxLength: 200
  });

  // Translate if needed
  const translation = await client.text.translate(
    summary,
    { targetLanguage: 'es' }
  );

  return {
    original: summary,
    spanish: translation
  };
}

Content Pipeline

async function contentPipeline(articles: string[]) {
  const results = [];
  
  for (const article of articles) {
    // Generate summary
    const summary = await client.text.summarize(
      article,
      { maxLength: 150 }
    );
    
    // Generate embedding for search
    const embedding = await client.embeddings.create({
      text: summary,
      model: 'embedding-v1'
    });
    
    results.push({
      summary,
      embedding: embedding.vector
    });
  }
  
  return results;
}