UA chatbot

Imagine having all the answers about the U of A ready for you at any moment, well here we present to you a AI assistant that can do just that!

Comment

Inspiration

We envisioned a chatbot that was all knowing about the University of Arizona, and would be able to provide guidance, help, or information to students with ease by simply visiting our site.

What it does

Our chatbot was designed to be able to answer almost any question related to U of A.

How we built it

Used Amazon AWS services starting with creating an S3 bucket that contained all of the data we scraped from U of A websites using Python. We then incorporated the S3 bucket into a knowledge base that acted as a data source for our AI agent to pull from.

Challenges we ran into

Unfortunately we weren't able to get a chatbot that functioned as smoothly as we had hoped. The root of our issues seem to have been due to either the data we scraped from the UA webpages or how our agent was utilizing the knowledge base.

Accomplishments that we're proud of

We were proud of being able to scrape over 350 UA webpages and collect plain text data from their HTML documents. We also are proud of being able to set up the initial components of an AI agent.

What we learned

It turns out to be quite the challenge to make a highly functional agent! But neither me or my teammate had used AWS before, and so that was a valuable learning experience and good introduction to just a small portion of the services AWS offers. We also used new Python libraries, 'requests' and 'beautifulsoup4' to scrap websites, parse data, and export the data to plain text files.

What's next for UA chatbot

Getting our chatbot to actually be competent! We'll probably need to do more work on improving our data and its structure. Then do more research on how to correctly build a AI agent. We also originally planned to have a webpage hosted on AWS that students could visit to talk with our chatbot.

Built With

Share this project:

Updates