Inspiration

This was our First Hackathon (Beginners), so we wanted to start our hackathon journey in a meaningful way. We were inspired by the recent arrival of ChatGPT plugins, especially the Wolfram Alpha plugin that circulated on social media. However, we were bummed when we found out that we wouldn't be able to experiment with plugins since it requires registering beforehand. This led us to think about using ChatGPT in another way.

What it does

Our project is a proof-of-concept of an application that has a content creator enter a YouTube link to one of their videos. It then parses the transcript for the YouTube video and uses ChatGPT to determine what questions might still be left unanswered for the role the user selected as well as questions answered in the video. The second feature of our application builds off the first, giving viewers the ability to find videos on specific topics and less-well known creators by matching their questions to videos which ChatGPT thinks answered those questions.

How we built it

We used a MERN stack for our application as well as the YouTube Transcript & ChatGPT APIs.

Challenges we ran into

Some challenges we faced included learning React, ChatGPT has a maximum number of tokens we can pass to it, extensive manipulation and filtering of the transcript data to convert it into a form of a closed caption. Overall, our biggest challenge was time, or lack thereof, for the scope of the project we wanted to complete.

Accomplishments that we're proud of

We're proud of the transformation of the transcript data into a usable form, as it was a significant, multi-step challenge that required great team collaboration and effort.

What we learned

We learned some React, how to work with the ChatGPT & YouTube Transcript APIs, and how to perform data manipulation.

What's next for YouTube Query

We are planning on finishing the application and then adding more features including allowing users to upvote or downvote questions and answers as well as suggesting sources for materials relevant to questions.

Share this project:

Updates