ZotNotes

Dashboard (lightmode)
Create New Course
AI Quiz Creation

Inspiration

As the next generation of students dependent on ChatGPT, we found ourselves at a loss when we realized it couldn't help us with textbook-specific content. We needed to be able to have an AI application in which textbooks can be read through and take in search queries to create study material, such as test questions.

What it does

ZotNotes allows users to log in to their respective Google account, which is linked to their ZotNotes profile. Here, they are able to add their current courses and their information onto their dashboard. They can search for their class-specific textbook and attach to it their course. From here, ZotNotes our AI pdf scraper scans and reads through a pdf, which sends the text information to our Open AI program to create a personalized quiz, depending on what the student requests. As students learn and get through their course, the progress bar for their course is updated on their dashboard. Ultimately, ZotNotes uses Artificial Intelligence as a resource that helps students keep track of their course study material and provides a tool that can help create personal practice quizzes.

How we built it

The source of all the textbooks is from the InternetArchive API, which only pulls the searched textbooks and saves them into our Firebase cloud. Here, the textbook URL collection is then accessed through the cloud and uses Langchain, Zilliz, and OpenAI through a Python program to do a semantic search through the textbook and make a quiz. We used React to bind the frontend, which is created with a variety of HTML, JS, and CSS scripts, to the backend with JSX scripts.

Challenges we ran into

As this was our first time using these kinds of APIs, it took us a long time to understand the documentation for the semantic search. We attempted using a variety of different APIs like Cohere, Langchain-community pdf readers, and Pinecone. Ultimately, we lost valuable time with our original attempt to implement Pinecone with our Langchain embeddings. We ended up landing on using Zillis and Langchain, which allowed us to execute our program correctly. Additionally, once the frontend and backend were finished separately, we struggled to merge the two together. As it was our first time using React, we spent a lot of time figuring out the syntax as well as integrating it with our Firebase cloud. We realized that our frontend was originally written in HTML and React only takes JSX files, which meant that we had to translate our entire frontend so it could be connected to our backend. Lastly, the Firebase cloud data had to be accessed through the frontend, back, and back to the frontend, we struggled with the syntax and logic but ultimately were able to execute our code.

Accomplishments that we're proud of

With all four hackers on our team being novice programmers and an incredibly ambitious idea of implementing a variety of AIs with a cloud dataset, we are proud of our final product. We were able to implement in 36 hours an interactive and clean-looking user interface, as well as, finishing implementing all of our APIs into the frontend, so a student would actually be able to ask for a mock quiz and respond back to the AI.

What we learned

Throughout the development of ZotNotes, we had to practically learn everything from the basics like HTML syntax, to more complex feats such as using React to connect our backend to our frontend with our Firebase cloud. Even though the entirety of ZotNotes was a learning curve for everyone, we all looked at every unknown aspect as a lesson, which we go to teach ourselves, ultimately building ZotNotes.

What's next for ZotNotes

When brainstorming for ZotNotes and what features we wanted to include, we came up with dozens of ideas. Unfortunately, with our limited time, we weren’t able to develop everything we wanted to. However, we are so excited to continue working on ZotNotes on our own time. In the future, we will be integrating UCI’s student login, which would allow us to have access to students' UCI id. Along with this, we will implement PetrPortal, we will be able to save students’ information and courses that they are taking this quarter. The only reason we weren’t able to is because the UCI OIT request would have taken 2 weeks to get permission from, which is time we didn’t have during the hackathon. With this, we would also be able to implement a calendar, in which students could see their courses as well as their homework, all in the same dashboard. We also will be adding a grade calculator based on the student’s performance with the AI quizzes that are generated.