Inspiration
We were inspired by observing the questions that were most frequently asked to TAs and academic advisors in the UW and realizing that many of these questions were duplicates and available on the university websites. By taking this information in and creating a bot powered by AI, we could reduce the amount of repeated questions and make information access easier to students.
What it does
Advisor Assistant is an AI chatbot that answers your frequently asked questions regarding UW admissions, class syllabi, and general UW related frequently asked questions.
How we built it
We built this bot using Microsoft Azure, webchat, and the QnAMaker service. For the front end, we used react.js and mui to make a clean yet effective front end
Challenges we ran into
A challenge we ran into was that this was the first hackathon experience for half of our group. This came with the territory that some of us were unfamiliar with azure, webchat, and react. For example, silly problems like npm start not running, or the code crashing because of having one line inserted was a challenge. In addition to that, we were on a time crunch, since the first half of the hackathon we were learning the materials that we were using.
Accomplishments that we're proud of
The main thing we are proud about is that we actually were able to make something! However, regarding just the bot we are proud that it can give helpful answers to frequently asked questions regarding UW. We are also proud that it uses machine learning and can improve its answer. Additionally, user interface looks professional and clean.
What we learned
We learned a lot about machine learning and how to use Microsoft azure. In addition, we learned how to use git with terminal, using npm, and using libraries for js. Specifically, making a bot, running it in a resource group, and setting up continuous deployment via github. Overall we learned what being a full stack developer entails and were able to experience it first hand.
Github Repos
What's next for Adviser Helper
For the frontend, we want to setup two views: the student view, and the advisor view. The student view being the chatbot front end you see now and the advisor view being a view were advisors are able to see unanswered questions and are able to respond to them. For the bot itself, we want to add a feature in the backend that allows the bot to automatically scrape the internet for updates FAQs on the UW website and update the bot’s knowledge base with an API call, instead of manually importing data for the bot to learn with. We also want to take document input from the user, such as a syllabus, so they won't have to read it.
Log in or sign up for Devpost to join the conversation.