bakalarka_praca/telegram_bot

Hate Speech Bot – Documentation

🧠 Application Overview

The Hate Speech Bot is a complete system for the automated detection of hate speech within Telegram. It functions as a monitoring and analysis tool that enables:

• Automatic interception and analysis of user messages in group chats
• When hate speech is detected, it records information about the user and the message into a database
• Displays violators through a real-time web interface

System Components:

1. Telegram Bot: Intercepts messages and uses a fine-tuned LLM model to classify whether the message contains hate speech.
2. MySQL Database: Stores user information, messages, and incident timestamps.
3. Web Application (Flask): Displays a real-time table of violations, updated every 5 seconds.

---

🧱 List of Used Docker Containers

Container	Description
tetianamohorian/hate-speech-bot	Contains the Telegram bot, classification model (e.g. fine-tuned BERT), and Flask frontend
mysql:8	Relational database for storing violator records

---

☸️ List of Kubernetes Objects

Object	Description
Namespace: botspace	Isolated space for all app-related objects
Deployment: bot-deployment	Deploys the container with the bot
Deployment: flask-web	Deploys the Flask web server
StatefulSet: mysql	Ensures persistent and consistent MySQL operation
PersistentVolume	Stores database data outside the pod
PersistentVolumeClaim	Requests disk space for the database
Service: mysql	Internal service for accessing the database
Service: flask-service	Port-forward service for accessing the web
ConfigMap: init-sql	Contains initial SQL schema scripts
Secret: bot-secret	Secure storage of sensitive data like the Telegram token

---

🌐 Virtual Networks and Storage

• All pods communicate over an internal network in botspace, minimizing data leakage risks
• PersistentVolume attached to MySQL ensures data preservation even after pod restarts

---

⚙️ Container Configuration

• Flask Web: Runs on port 5000, accessible via flask-service. Automatically updates via JavaScript (AJAX fetch) every 5 seconds
• Telegram Bot: Runs in a separate thread, classifies messages using a PyTorch model or HuggingFace Transformers, and logs results into the database
• MySQL: Initialized with an initContainer that runs a script from the ConfigMap

---

🚀 Usage Guide

1. Prepare the Application:

./prepare-app.sh

• Builds the Docker image
• Pushes it to Docker Hub

---

2. Start the Application:

./start-app.sh

• Creates Kubernetes objects
• Starts port-forwarding for Flask web on localhost:8888

---

3. Stop Port-Forwarding:

pkill -f "kubectl port-forward"

• Terminates temporary access via port-forward

---

4. Remove the Application:

./stop-app.sh

• Deletes the entire botspace, including database and services

---

🌍 Accessing the Web Interface

After launching the app, open a browser:

http://localhost:8888

• A table of violators will be displayed
• Data updates automatically every 5 seconds

---

📦 Technologies Used

• Python, Flask, PyTorch / Transformers – text analysis and classification
• Docker, Docker Hub – containerization
• Kubernetes – orchestration, scaling, and network isolation
• MySQL – relational data storage
• JavaScript (AJAX) – real-time frontend updates