Inspiration / Overview

We created a chatbot platform that recreates your personality based on your online social interactions. Not only do we study human interactions, but also provide a chatbot framework geared toward machine learning research.

What it does, a novel machine learning platform, allows you to create versatile and scalable chatbots that are compatible on a bunch of platforms. Tinder? Twilio? Done and done. Our chatbots will maintain your conversations for you.

How we built it uses your Facebook Messenger corpus to learn about you. We train our TensorFlow ML model using this data for our word2vec and sequence2sequence algorithms. We use the IBM Watson API to implement tonal analysis and direct the flow of the conversation just as you would. We use the Tinder and Twilio APIs to set up bots to maintain your conversations on these platforms. We also use the API to manage our container services.

Challenges we ran into

Training our ML model took a LOT of time due to a lack of physical computing power. Interfacing through Tinder and debugging our Flask servers consumed a lot of time, too.

Accomplishments that we're proud of

Creating a platform that talks for you.

What we learned

Testing our implementations on Tinder is pretty entertaining.

What's next for

Making a modular implementation of this chatbot service.

Share this project: