NANNY

🧱 Technical Specification: Humor Analysis System Based on TikTok Likes

A calibrated project, with a spark. Not some bullshit.

💥 Overall Concept:

The bot operates in Telegram. A user enters, specifies their TikTok username (without password, only public username). The bot retrieves their liked videos, analyzes what types of videos they like, and provides a humorous profile — showing what makes them laugh: a girl with a brick, a raccoon in a microwave, or a dad shouting from the balcony.

Next — based on such profiles, a matching algorithm will work: finds other people with similar humor. This is the basis for dating through humor.
Genius? Yes. Difficult? Yes. Feasible? Also yes.

🔧 What the bot should do:

1. Interface (Telegram bot)

• /start — displays a button: “Specify TikTok username”

• User clicks — enters username

• Bot saves the username, says: “🟢 Username saved: e1xh. Now press ‘Get Likes’”

• Button “Get Likes” — user presses, bot retrieves liked videos

• ⚠️ Important: bot warns that likes must be public, otherwise it won’t work.

2. Parsing TikTok Likes

• Uses the library TikTokApi version 7.1.0

• Bot logs in anonymously via saved tiktok_cookies.json

• Gets a list of user’s likes (up to 100, can be more) by username

• If no likes or profile is private — informs the user, does not crash

3. Content Analysis

• Extracts metadata from each video:

  • description

  • hashtags

  • number of likes, comments

  • author

  • sound

• Runs through a simple NLP model or classifier:

  • humor = cringe / meme / life / random / joke / aggression / drug influence

  • saves to database (SQLite for now)

4. Profile Formation

• Builds a “humorous profile” based on all likes

• For example:

  arduinoCopyEditYour comedy is 40% cringe, 30% memes, 20% "life", 10% drug influence.
  You are the type who laughs when someone falls down the stairs.
  

• Displays to the user + saves in the system

5. Matching Selection

• Initially, just store everyone who passed analysis

• When there are 2+ users — perform profile comparison

• If 80% match — suggest user as “👀 Funny Match”

🗂️ What you as a developer need to know:

• Python 3.11+

• python-telegram-bot v20+ (works with asyncio)

• TikTokApi==7.1.0 + cookies file

• Telegram Bot Token is already available

• UI in the bot — via InlineKeyboardMarkup

• Store data locally (SQLite or JSON, if lazy)

• Later, a web interface can be added, but for now — Telegram

❗ Important details:

• Remember — TikTok restricts: you need to use cookies, and likes must be public

• Videos do not need to be downloaded — only metadata

• Preferably, parsing and analysis should work asynchronously, without freezing the bot

🧱 The project is a prototype, but export-ready. Not just a “bot,” but a database for neural matchmaking based on humor.
If done properly — it can later be scaled into a separate service.