message-analyser

Statistical analysis of VKontakte and Telegram message history.

Dependencies

• Telethon
• seaborn
• wordcloud

Installation

• Use Python3.6. 3.7 version may not work properly.
• git clone https://github.com/vlajnaya-mol/message-analyser
• Install requirements.txt. (pip install -r /path/to/requirements.txt)

Usage

Execution

• Run python main.py
• Follow GUI commands

Telegram messages

• You need API Hash and API ID from here
• Be sure to use proxy if Telegram is blocked in your country.

VKontakte messages

• Install VkOpt extension

• Save Your conversation as .txt file using this extension

  Be sure You used default format settings:

  %username% (%date%):
  %message%
  
  HH:MM:ss  dd/mm/yyyy

• You can concatenate two VkOpt files and use as one
• Include this file in the analysis process

Words

• Write words You are interested in to a file
• Be sure words are saved correctly. Cyrillic words are ruined by saving in ASCII format.
• Include this file in the analysis process

Manual analysis

• Fill config.ini file and use retrieve_and_analyse(loop) instead of using GUI.
• Use analyse_from_file(path) function instead of redownloading messages

Examples

• All examples can be found here
  
  
  
  
  
  
  
  

Potential project improvements

• analysis of group chats.
• improve tkinter theme.
• add VkOpt stickers as emojis to messages.
• add plot correlation between the number of voice messages and the average message length.
• add "first-to-write" and "response time (delay)" plots (lineplot).
• add n-grams plot (lineplot).