This project was built as a live demo for a project in my CSC485E (Intro to Natural Language Processing) class at UVic. My team built a framework to train multiple classifiers and run examples efficiently.
The models were built around Train/Predict functions in which we trained each on a CSV file found here: https://www.kaggle.com/tovarischsukhov/southparklines. We used the NLTK library to do our heavy lifting.
The brains for this site were deployed on my VPS. It’s just a little Python Flask app running on Apache mod_wsgi. Our models were pre-trained locally and pickled. Each character/model pair has its own pickle to make things a bit more manageable. I’ll be outlining the challenges of this deployment choice along with my solutions in my blog soon.
The front end is just simple HTML/JS/CSS with a little parallax bonus because parallax is so 🔥 right now.
We’re working on improving our models while we finish this project for August. Have fun and enjoy!Back to albums