Dating App Prototype

Description

The Dating App Prototype is a proof-of-concept matchmaking system that uses personality traits and embeddings to find compatible matches. The system relies on MBTI personality types, Big Five personality traits, and vector embeddings to calculate match scores between users. The prototype is built with Flask and integrates simple authentication, data storage, and matchmaking functionality.

Matching Algorithm and Weights

The matchmaking system uses the following weighted criteria to calculate compatibility:

• MBTI Compatibility (20%): Uses a predefined compatibility chart for all 16 MBTI types.
• Big Five Personality Traits (40%): Calculates similarity using Euclidean distance between users' Big Five traits.
• Hobbies Similarity/ Vector Embeddings (40%): Matches users based on common hobbies Using cosine similarity between user embeddings generated from provided data.

Overview

This prototype is a minimal implementation designed for testing and development. The app allows users to sign up, input personality traits, and check their match scores with other users. The backend is implemented using Flask, and data is stored in JSON files.

File Structure

/Dating-app-prototype
  ├── app.py
  ├── venv
  ├── .env
  ├── .gitignore
  ├── generate_embeddings.py
  ├── generate_fake_data.py
  ├── match_scores.json
  ├── matchmaking.py
  ├── requirements.txt
  ├── user_embeddings.json
  ├── fake_user_data.json
  └── templates/
      ├── home.html
      ├── login.html
      ├── sign_up.html
      ├── form.html
      ├── profile.html
      └── test.html

File Descriptions

• app.py - The main Flask application that handles user authentication, form submissions, and rendering templates.
• generate_embeddings.py - Generates vector embeddings for users based on their personality traits.
• generate_fake_data.py - Creates a set of fake user profiles for testing purposes.
• matchmaking.py - Runs the matchmaking algorithm, calculating match scores based on MBTI, Big Five traits, and vector similarity.
• match_scores.json - Stores the match scores for each user after running the matchmaking script.
• user_embeddings.json - Stores the embeddings for each user, which are used for similarity calculations.
• fake_user_data.json - Stores user profiles, including personality traits and preferences.
• templates/ - Contains HTML templates for the frontend.
  • home.html - Main landing page.
  • login.html - User login page.
  • sign_up.html - User registration page.
  • form.html - Profile update form.
  • profile.html - Displays user profile.
  • test.html - Displays match results.
  • chat.html - Chat UI where users can message with matches users. (Issue: both users don't need to accept)
  • accepted_rejected.json - Store the accepted and rejected user choices.
  • messages.json - Stores all the messages of users.

Setup and Installation

1. Clone the repository:

   git clone https://github.com/your-username/Dating-app-prototype.git
   cd Dating-app-prototype

2. Create a virtual environment and activate it:

   python -m venv venv
   On Windows use `venv\Scripts\activate # mac 'source venv/bin/activate'

3. Install dependencies:

   pip install -r requirements.txt

4. Create .env file and add the OpenAI API and Your Deepseek API keys:

   OPENAI_API_KEY=sk-proj-XXXXXXXXXXXXXXXXXXXXXXX
   DEEPSEEK_API_KEY=XXXXXXXXXXXXXXXXXXXXXXXXXXXXX

5. Run the Flask application:

   python app.py

6. Open your browser and go to:

   http://127.0.0.1:5000/
   

You can run these files mannualy for testing purposes

1. Create user embeddings:

   python generate_embeddings.py

2. Run the matchmaking script:

   python matchmaking.py user_0  # Replace `user_0` with the actual user ID (eg- user_1 , user_6)

3. View the matches in test.html.

You can run these files to generate fake user data

1. Create fake users:

   python generate_fake_data.py

Security Concerns

• No password hashing: User IDs are used as passwords, which is insecure for real-world applications.
• No database: All data is stored in JSON files, which can be easily modified.
• No user session management: Users are not properly authenticated beyond checking IDs.
• Vulnerable to data corruption: No error handling for file modifications.

Future Improvements

• Implement proper authentication with hashed passwords.
• Replace JSON storage with a database (e.g., PostgreSQL, MongoDB).
• Improve the matchmaking algorithm using more robust ML techniques
• Enhance frontend UI for better user experience.

This project is for testing purposes only and is not intended for production use.