Wingman

Wingman is your perfect companion to creating an excellent Tinder profile. Using advanced machine learning techniques, Wingman scrapes through tens of thousands of Tinder profiles for both males and females and then generates sample Tinder bios. With Wingman, you'll never have a boring profile again!

What We Used

Wingman was built using Tinder Scraper and Torch-RNN. The web client was built using HTML, CSS, Javascript and Bootstrap. Python was used to write the scripts to simplify the data. After gathering data from 15,000+ people, we used the Long Short Term Memory Algorithm to have a machine learn from the data and output random bios.

Future Updates

Due to the time constraint and computational power limitation, we could have the website to work on generating female bios. But given more time and data, we can also implement it for males. We also wanted to host the data on a database which the website can access for future use. Extra options such as a creativity scale can be implemented to create a bio to the user's liking. We can expand beyond the bios and work with images the user imports. Based on the data given, we can analyze it if it a suitable picture to post. In the end, however, training the neural network more would create not only more realistic bios, but also give rise to more functionality.

+ 4 more
Share this project:

Updates