[35] - Quizzify

Quizzify home page
questions

Inspiration:

As a person with neurodivergent disorders I struggle to pay attention and focus on information about lectures during and after the lecture. I want a way to help me learn and pay attention better during lectures. I wanted an engaging way to keep me focused and motivated during lectures.

What it does

Quizzify is designed to generate comprehensive questions during a lecture. Simply start recording during the lecture and it will start generating quiz questions. At any point when you would like to check your knowledge, stop the recording and test your knowledge of the content covered so far on the website. It also allows you to generate questions after a lecture by taking in the lecture transcript and generating questions based on what was taught in the lecture.

How we built it

We built the frontend using Next.Js, Chakra UI, and React, while using Python in the backend. We used various machine learning models to create our own type of machine learning model to generate questions based on the transcript, create fake answers as answer choice, and using an online api to help get synonyms to feed into our models.

Challenges we ran into

Understanding the machine learning that we are using and time complexity. Often running big machine learning models takes time so trying to analyze the text and transcript that is being sent in takes time, and so trying to optimize that is very difficult. Connecting the front end to the backend was very difficult, because we had to play with different ways of sending information to the backend and then in the backend we had to figure out what format to put the information in to send to the frontend to make it easiest to display the questions.

Accomplishments that we're proud of

We are proud that we were able to figure out the various python libraries and how to use machine learning, without using the openAi api to generate our questions. We are proud that we were able to deploy the frontend and tunnel the backend. We are proud that our members were all able to work on different parts of the application and learn a lot throughout the process.

What we learned

We learned how to use Hugging Face and BERT to use machine learning to do various tasks based on language. We started to understand better how to use machine learning models and how to integrate that with our own python code to optimize various processes in the backend. In the frontend we learned how to better use Chakra UI and Next.Js for development purposes. We also learned how to generate pages based on information fed in from the backend.

What's next for Quizzify

Next for Quizzify is optimizing the model for better questions so that students can learn more and better test their knowledge. We want to implement audio recording so that students can quiz and test themselves during class to stay engaged and make sure they are retaining the material they are learning.