Our group agreed we wanted use text classification for a data set that is not used often. There aren't many wine recommenders out there or ones we could think of off the top if our heads, so we built one ourselves.
What it Does
This is a recommendation system where a user can describe the type of wine they enjoy and the our system will present the user with varieties of wine they like best, based off their characteristics.
How it was Made
All the initial scripts and testing were done in Jupyter before transferred to actual python programs. React was used so that a user can test the text classification by typing in a sentence explaining their wine preference.
Challenges we ran into was setting up the Bert server, leaning data frames, and running into pandas. I ran into problems in react when I tried to integrate the data set into the search box.
I am proud that I learned how to use react in only a couple hours and gain general knowledge on machine learning and natural language programming.
We could take this project further by using other databases.