AI-Powered PDF Keyword Search & Classification Tool

Overview

This project is a Flask-based web application that allows users to upload PDF documents and search for specific keywords. The tool provides two search modes:

1. Regular Search – Finds and displays exact keyword matches.
2. Smart Search – Uses OpenAI's GPT API to determine whether the keyword appears in a context relevant to health or climate.

The system outputs:

• The pages where the keyword appears.
• The sentences containing the keyword.
• A classification of whether the sentence is contextually relevant to health or climate.

Features

✅ Upload PDFs (Max size: 16MB)
✅ Extract and search text from PDFs
✅ Keyword matching with contextual classification using GPT-3.5 Turbo
✅ User-friendly web interface with Flask and HTML/CSS
✅ Deployment-ready (Heroku or local execution)

---

Tech Stack

• Python (Flask, pdfplumber, re, dotenv, OpenAI API)
• Frontend: HTML, CSS, JavaScript
• Cloud Deployment: Heroku (or can be run locally)

---

Installation & Setup

1️⃣ Clone the Repository

 git clone https://github.com/your-repo-url.git
 cd your-repo-folder

2️⃣ Create a Virtual Environment (Optional but Recommended)

 python3 -m venv venv
 source venv/bin/activate  # On Windows: venv\Scripts\activate

3️⃣ Install Dependencies

 pip install -r requirements.txt

4️⃣ Set Up OpenAI API Key

Create a .env file in the project root and add your OpenAI API key:

 OPENAI_API_KEY=your_openai_api_key_here

5️⃣ Run the Application

 python app.py

6️⃣ Open in Browser

Once running, open http://127.0.0.1:5000/ in your web browser.

---

Usage

Uploading & Searching:

1. Upload a PDF file.
2. Enter comma-separated keywords in the search box.
3. Select search mode: Regular or Smart Search.
4. Submit to process the document and view results.

Results Page Includes:

• Found Keywords: Page numbers and matched sentences.
• Context Analysis: Indicates whether the keyword is used in a health/climate-related context.
• Download/Review: Ability to go back and upload another file.

---

File Structure

📂 project-root
 ┣ 📂 templates/          # HTML Templates (upload.html, results.html)
 ┣ 📜 app.py              # Main Flask Application
 ┣ 📜 requirements.txt    # Dependencies
 ┣ 📜 .env.example        # Sample environment variables file
 ┗ 📜 README.md           # Project Documentation

---

Future Improvements

🔹 Support for additional document formats (DOCX, TXT)
🔹 Multilingual support for keyword classification
🔹 Improved UI/UX with React or Vue.js

---

License

MIT License. Feel free to use and improve!

---

Contact

For inquiries, contact Primanta at primanta.b@columbia.edu.