🧠 AI Fatwa QnA Generator

An AI-powered automation project that generates question–answer pairs from Islamic text datasets using Large Language Models (LLMs).
It integrates Hugging Face models (Mistral / Falcon) and automatically falls back to OpenAI GPT-3.5 Turbo if needed — ensuring reliable, consistent output even when certain APIs are restricted.

---

🚀 Overview

🎯 Objective

To demonstrate an end-to-end AI automation workflow:

• Extract Islamic fatwa or Q&A content from structured text/Excel.
• Generate meaningful questions and concise answers in Bahasa Melayu.
• Use multiple AI model endpoints with built-in fallback handling.

---

🧩 Tech Stack

Category	Tools / Frameworks
Programming	Python 3.13
LLM APIs	Hugging Face (Mistral-7B, Falcon-7B), OpenAI GPT-3.5
Data Handling	pandas
Automation	dotenv (API key management), Playwright (optional web scraper)
Environment	Visual Studio Code + PowerShell
Dataset	irsyad_fatwa_umum_all_pages.xlsx (Fatwa / Islamic content)

---

⚙️ Setup Instructions

1. Clone or copy this folder to your local environment: AI_Fatwa_QnA_Generator/

2. Create a .env file in the project root: OPENAI_API_KEY=your_openai_key_here HF_API_KEY=your_huggingface_token_here

3. Install dependencies: pip install -r requirements.txt

4. Run the Q&A generator: python demo_generate_qna.py

---

🧠 Example Output

✅ Excel file loaded successfully. Columns: ['E', 'F', 'I'] ✅ Using text from column: 'E'

🧠 Generating questions... ⚠️ All Hugging Face models unavailable – falling back to OpenAI GPT-3.5... ✅ OpenAI responded successfully.

❓ 1. Apakah hukum berkaitan penggunaan produk Trend Body Tea dalam perspektif Fiqh ❓ 2. Bagaimanakah Trend Body Tea mempengaruhi kesihatan seseorang dari sudut pandangan agama ❓ 3. Adakah terdapat panduan dalam Fiqh mengenai penggunaan produk seumpama ini

💬 Hukum penggunaan produk Trend Body Tea dalam perspektif Fiqh adalah harus selagi tidak mengandungi bahan yang diharamkan dalam Islam.

---

🧱 Project Structure

AI_Fatwa_QnA_Generator/ │ ├── demo_generate_qna.py # Main script for testing QnA generation ├── mufti_irsyad_scraper_playwright.py # Web scraper for Islamic fatwa data (optional) ├── llm_utils.py # Core logic: HuggingFace + OpenAI integration ├── requirements.txt # Project dependencies ├── .env # API keys (excluded from repo) ├── irsyad_fatwa_umum_all_pages.xlsx # Dataset for testing └── README.md # Project overview (this file)

---

💡 Future Improvements

• Add multilingual support (Arabic → Malay translation).
• Integrate with Streamlit for a simple web interface.
• Export generated QnA pairs into JSON or database for chatbot training or retrieval tasks.

---

👤 Author

Ahmad Syauqi Bin Amran Nor syauqiamran@yahoo.com LinkedIn Data Analyst | AI & BI Enthusiast
Kuala Lumpur, Malaysia

---