Connecting-the-Dots-Round-1A-and-Round-1B-

Adobe PDF Intelligence Challenge

This repository contains solutions for the Adobe Hackathon rounds focusing on PDF document intelligence. It includes implementations for:

• Round 1A: PDF Outline Extractor — Extracts structured outlines (title and hierarchical headings) from PDF files (≤ 50 pages), fully offline, using PDF.js.
• Round 1B: Persona-Driven Document Intelligence — Analyzes a collection of PDF documents based on a specified persona and job-to-be-done, ranking and extracting the most relevant sections client-side.

---

Table of Contents

• Overview

• Round 1A: PDF Outline Extractor

  • Description
  • Tech Stack
  • Installation & Usage
  • Project Structure

• Round 1B: Persona-Driven Document Intelligence

  • Description
  • Tech Stack
  • Installation & Usage
  • Project Structure

• Contributing

• License

---

Overview

Adobe’s "Connecting the Dots Challenge" aims to reimagine PDF interaction by enabling automated structure extraction and persona-driven insights. This repo implements client-side solutions for both document outlining and intelligent section extraction, paving the way for richer, more context-aware PDF experiences.

---

Round 1A: PDF Outline Extractor

Description

Builds the "brains" of the challenge by extracting a clean, hierarchical outline (Title, H1, H2, H3) from any PDF up to 50 pages. Results are rendered and available for download as JSON.

Tech Stack

• HTML5 & CSS3 for layout and styling
• Vanilla JavaScript (ES6) for application logic
• PDF.js for offline PDF parsing

Installation & Usage

1. Clone the repository:

   git clone <repo-url>
   cd <repo-root>/round-1A

2. Open index.html:

   • Simply open round-1A/index.html in your browser (no server required).

   • Or launch with VS Code Live Server:

     1. Install the "Live Server" extension.
     2. Right-click index.html → Open with Live Server.

3. Upload or drag & drop your PDF (≤ 50 pages, ≤ 10 MB).

4. Click Extract Outline to generate and download the JSON result.

Project Structure

round-1A/
├── index.html       # Main UI
├── style.css        # Styling and responsive layout
└── app.js           # Core logic: validation, PDF.js integration, outline generation

---

Round 1B: Persona-Driven Document Intelligence

Description

Processes 3–10 PDF documents against a specified persona and job-to-be-done. Scores and ranks sections by keyword relevance, returning the top insights in JSON format.

Tech Stack

• HTML5 & CSS3 for basic UI
• Vanilla JavaScript (ES6 modules)
• PDF.js MJS for in-browser PDF parsing

Installation & Usage

1. Clone the repository:

   git clone <repo-url>
   cd <repo-root>/round-1B

2. Serve the files: Modern browsers enforce module CORS rules, so serve via a static server:

   npx http-server .

   or use VS Code Live Server.

3. Navigate to the served URL (e.g., http://127.0.0.1:8080).

4. Upload 3–10 PDFs, enter Persona and Job to be‑done, then click Analyze Documents.

5. View or copy the generated JSON output.

Project Structure

round-1B/
├── index.html       # UI form for file upload and inputs
├── styles.css       # Basic styling for form and output
└── app.js           # Logic: text extraction, keyword scoring, JSON formatting

---

Contributing

Contributions are welcome! Please open issues or pull requests to improve functionality, fix bugs, or enhance documentation.

---