ContentAnalyzer Chrome Extension

A Chrome extension for processing and analyzing government forms of any type. The extension extracts form data from PDFs, converts it to a structured digital format, and displays it in a user-friendly interface with modern UI components. It also provides powerful text analysis capabilities for web content.

Features

• Extract form data from any PDF forms without hardcoded templates
• Process multi-page PDF documents with consistent results
• Display form data in a large, responsive viewer window
• Support for complex nested objects and field types
• Save, edit, print, and export form data
• AI-powered field recognition and categorization
• Sentiment analysis for web page content
• Text processing from multiple sources (page context, clipboard, selected text)

Architecture

The ContentAnalyzer implements a sophisticated multi-agent architecture for form processing, consisting of specialized agents that work together in a pipeline:

Multi-Agent Form Processing Architecture

1. Parsing Agent

   • Specialized in OCR and text extraction from forms
   • Understands document structure and layout
   • Identifies form fields, labels, and values with high accuracy
   • Processes multi-page documents and complex layouts

2. Builder Agent

   • Transforms raw extracted data into structured form objects
   • Validates field values against expected formats
   • Resolves ambiguities in field identification
   • Groups related fields into logical sections

3. Designer Agent

   • Creates user-friendly form layouts from structured data
   • Implements multi-step navigation for complex forms
   • Optimizes field arrangement for better usability
   • Applies appropriate UI components for different field types

4. QA Agent

   • Verifies extraction accuracy and completeness
   • Identifies potentially missing required fields
   • Checks for inconsistencies in the extracted data
   • Provides confidence scores for the overall form processing

This modular approach provides several benefits:

• Each agent can be improved independently
• Specialized AI models can be used for specific tasks
• New form types can be supported without modifying the entire system
• QA validation ensures accuracy and completeness

Usage

Processing Forms

1. Click the ContentAnalyzer extension icon in Chrome
2. Upload a PDF form document
3. The extension will automatically:
   • Process all pages in the document
   • Extract form fields and their values
   • Categorize fields into logical sections
   • Present the form in a user-friendly layout

Form Viewer Features

• Search: Find specific fields quickly with the search function
• Zoom Controls: Adjust text size for better readability
• Edit Mode: Make changes to extracted form data
• Print/Export: Print or export the processed form
• Field Navigation: Easily navigate between form sections

Working with Complex Fields

The system handles various field types with specialized UI components:

• Text Fields: Standard input fields for text
• Date Fields: Calendar-based date pickers
• Checkboxes/Radios: Toggle fields for boolean values
• Select Fields: Dropdown menus for option selection
• Complex Objects: Expandable objects with multiple properties
• Arrays: List-based fields with add/remove functionality

Sentiment Analysis and Text Processing

ContentAnalyzer also provides powerful text analysis capabilities:

Sentiment Analysis

Analyze the sentiment of text from various sources:

• Selected Text: Right-click on any selected text on a webpage to analyze its sentiment
• Page Content: Analyze the sentiment of an entire webpage's content
• Custom Input: Paste or type text directly for sentiment analysis

The sentiment analysis provides:

• Overall sentiment score (positive, negative, or neutral)
• Confidence level for the sentiment determination
• Key phrases that influenced the sentiment score
• Emotional tone detection (joy, anger, sadness, etc.)

Text Sources

The extension can process text from multiple sources:

• Page Context: Automatically extract and analyze content from the current webpage

  • Detects main content areas vs. navigation/ads
  • Identifies relevant sections based on context
  • Handles dynamic content loading

• Clipboard Access: Process text directly from your clipboard

  • Right-click context menu option for "Analyze Clipboard Content"
  • Supports formatted text with structural preservation
  • Handles both plain text and rich text formats

• Selected Text: Process specifically highlighted portions of text

  • Maintains context from the source document
  • Supports partial selection from larger documents
  • Provides context-aware analysis based on surrounding content

Usage Examples

• Analyze sentiment of product reviews on e-commerce sites
• Extract key information from news articles
• Summarize long documents by selecting important passages
• Verify emotional tone of your own writing before sending

Installation

1. Clone the repository
2. Install dependencies with npm install
3. Build the extension with npm run build
4. Load the extension in Chrome from the dist directory

Development

Setup

npm install
npm run dev

Build

npm run build

Using the Multi-Agent System in Code

To process forms programmatically:

import { createFormProcessingPipeline } from './src/lib/agents';

// Create the pipeline
const pipeline = createFormProcessingPipeline();

// Process a form with PDF pages
const processForm = async (pdfPages) => {
  const result = await pipeline.processForm(pdfPages, {
    formNumber: 'FORM-1234', // Optional
    formTitle: 'Sample Form', // Optional
  });
  
  // Access the processed form data
  console.log(result.formData);
  
  // Check processing confidence
  console.log(`Confidence: ${result.confidence}`);
  
  // View any issues found during processing
  console.log(result.issues);
};

// Or run a specific stage of the pipeline
const runParserOnly = async (pdfPages) => {
  const parserResult = await pipeline.runStage('parser', {
    pages: pdfPages,
    mode: 'detailed'
  });
  
  console.log(parserResult);
};

Text Analysis API

// Sentiment analysis API
import { analyzeSentiment } from './src/lib/sentiment-analyzer';

// Analyze text from different sources
const analyzePage = async () => {
  const result = await analyzeSentiment({
    source: 'page',
    url: window.location.href
  });
  console.log(result.sentiment, result.confidence, result.keyPhrases);
};

const analyzeSelection = async (selectedText) => {
  const result = await analyzeSentiment({
    source: 'selection',
    text: selectedText,
    context: document.title
  });
  console.log(result.sentiment, result.emotionalTones);
};

const analyzeClipboard = async () => {
  const clipboardText = await navigator.clipboard.readText();
  const result = await analyzeSentiment({
    source: 'clipboard',
    text: clipboardText
  });
  console.log(result);
};

Technologies

• React
• TypeScript
• Vite
• PDF.js
• OpenAI API
• Tailwind CSS
• shadcn/ui components

License

MIT License