🛡️ Cybersecurity RAG Chatbot

Introduction

A context-based chatbot designed to answer specialized cybersecurity questions based on 500+ scientific cybersecurity articles (in text format). It provides quick, fact-based answers – solely grounded in the available documents.

---

📌 Project Team

This project was created as part of a module at the Bern University of Applied Sciences by:

• Carlos Gomez
• Nelson Vidovic
• Lirim Zahiri
• Mouad Medini
• Niklaus Joel

---

✅ Prerequisites (one-time setup)

1. 🔧 Install Python (version 3.9 or higher)

• Official website: https://www.python.org/downloads/
• During setup, enable "Add Python to PATH"
• Then verify installation in your terminal:

python --version

2. 🧑🏽‍💻 Create a Together.ai Account

1. Go to https://www.together.ai.
2. Click "Get started" and register (no credit card needed).
3. After logging in, your API KEY will show up automatically.
4. If not, go to your dashboard at https://api.together.ai.
5. Navigate to "API Keys" (bottom-left corner).
6. Click "Create new API key", give it a name (e.g. chatbot-key), and copy the key.
7. Save the key securely.
8. This key is free, but subject to rate limits (e.g. ~50 requests/min, ~200,000 tokens/day).

---

🔧 Setup – Step by Step

!For the following steps use CMD (Windows) / Terminal (Mac)!

1. 📁 Clone the Project

git clone https://github.com/VNel/SAI3_Chatbot.git
cd SAI3_Chatbot/online_chatbot

2. 🪪 Set Up Virtual Environment

• Windows:

py -m venv venv

• macOS:

python3 -m venv venv

Then activate:

• Windows:

venv\Scripts\activate.bat

• macOS:

source venv/bin/activate

---

3. 📦 Install Required Dependencies

pip install -r requirements.txt

requirements.txt content:

fastapi
uvicorn
requests
gradio
python-dotenv
faiss-cpu
sentence-transformers
numpy

---

4. 🔐 Add Your API Key

1. Select the file .env.template
2. Rename it to .env
3. Then open .env and insert your API key:

TOGETHER_API_KEY=<your_api_key>

Without .env, the chatbot cannot communicate with Together.ai.

---

5. 📚 Prepare Vector Database

• Windows:

py load_data.py

• macOS:

python3 load_data.py

• Creates documents.json (text chunks)

• Windows:

py build_index.py

• macOS:

python3 build_index.py

• Creates index.faiss (semantic search index)

---

6. 🚀 Launch Web Interface

• Windows:

py web_ui.py

• macOS:

python3 web_ui.py

Automatically opens: http://localhost:7860

---

💬 Features

• Ask questions in German or English
• Answers are based only on local documents
• No hallucination
• Transparent source citation for each answer
• If no relevant info is found → clear feedback

---

🔒 Security & Privacy

Item	Status
API key remains local	✅
No cloud storage	✅
No login required	✅
Local access via localhost	✅

---

✅ All Set! – Functionality Checklist

Requirement	✅
Python 3.9 or higher installed	✅
Virtual environment active	✅
pip install successful	✅
.env with API key present	✅
load_data.py ran successfully	✅
build_index.py successful	✅
web_ui.py started	✅

---

📂 Folder Structure (excerpt)

├── clean_data/               # 500 cybersecurity articles (.txt)
├── vectorstore/             # Vector DB (auto-generated)
│   ├── documents.json
│   └── index.faiss
├── .env                     # (created by you)
├── load_data.py             # Process texts
├── build_index.py           # Create FAISS index
├── query_pipeline.py        # Retrieval logic
├── llm_api.py               # Together.ai API integration
├── web_ui.py                # Launch web interface
├── requirements.txt         # Python dependencies
└── README.md                # This guide

---

📌 Note

This project is solely intended for research and educational purposes within the Business Information Technology program. It is not intended for production or commercial use.