Adobe India Hackathon 2025

Welcome to the "Connecting the Dots" Challenge

This repository contains the solution for the Adobe India Hackathon 2025, including the persona-driven document intelligence system for Challenge 1b.

Challenge 1b: Persona-Driven Document Intelligence

This solution is a Python-based application that analyzes a collection of PDF documents and extracts the most relevant sections based on a given persona and job-to-be-done.

Features

• Persona-Based Analysis: Ranks document sections based on their relevance to a user's persona and task.
• Extractive Summarization: Provides concise summaries of the most important sections.
• Dockerized Solution: The application is containerized for optional deployment and execution.

Project Structure

.
├── Dockerfile
├── README.md
├── approach_explanation.md
├── requirements.txt
├── src/
│ ├── main.py
│ ├── pdf_parser.py
│ ├── ranking.py
│ └── summarizer.py
└── Challenge_1b/
    └── [Collection Name]/
        ├── PDFs/
        │   ├── document1.pdf
        │   └── ...
        └── challenge1b_input.json

⚙️ How to Run the Solution

You can run the solution with or without Docker.

---

✅ Recommended: Run Locally Without Docker (CPU + Offline Compatible)

This solution is fully functional on CPU and does not require an internet connection once dependencies are installed.

1. Install Dependencies (Manually):

   Important: Installing spacy inside Docker can be time-consuming. To avoid long build times, it is highly recommended to install all packages separately on your system before execution.

   Run the following command in the root directory:

   pip install -r requirements.txt

2. Download Required Spacy Model (only once):

   If you're online during setup, run:

   python -m spacy download en_core_web_sm

   💡 Skip this if the model is already downloaded or if running offline.

3. Prepare Input Data:

   • Inside the Challenge_1b directory, create a new folder for your collection.
   • Add a PDFs/ subdirectory with all your input PDFs.
   • Add a challenge1b_input.json file at the root of that collection folder (use examples as reference).

4. Run the Application:

   From the root project directory:

   python src/main.py

   This will process all collections under the Challenge_1b directory and generate corresponding challenge1b_output.json files inside each collection folder.

---

🐳 Optional: Run with Docker (if needed)

1. Build Docker Image:

   ⚠️ Note: Building the Docker image may take time due to Spacy installation.

   docker build --platform linux/amd64 -t document-intelligence .

2. Run the Docker Container:

   docker run --rm -v "%cd%/Challenge_1b":/app/Challenge_1b document-intelligence

   This will process all collections and generate the output files in the same directory.

---

📄 For More Details

Refer to approach_explanation.md for a full explanation of the methodology, design, and component structure.