YouTube has lots of high quality educational content produced by dedicated channels such as 3Blue1Brown and Kurzgesagt. However, you're stuck with a malicious recommendation algorithm that tracks your every click and with the only purpose to make you fall down rabbit holes and offers distractions at every step. This makes it problematic for schools to use and for students to use as a study tool.

We present EduBytes, an alternative front-end for YouTube with a distraction-free user interface focused on high-quality, curated educational videos with a built-in recommendation system that respects your privacy and giving you full control over your data.

What it does

  • Has a search engine that gives relevant videos
  • Each video has tags improved from those directly obtained from youtube videos
  • Recommends videos based on search history

How we built it

  • Video Curation - Over 1000 video details scraped from popular educational channels using the YouTube Data API and adding tags via Python NLP libraries.
  • Frontend - Typescript with React.JS as a single-page application using local storage to keep your history and recommendations.

Challenges we ran into

  • Javascript Promises and inconsistent API documentation
  • YouTube Data API rate limits
  • Typescript integration for untyped libraries
  • github.com being blocked by Eduroam in the middle of the night

Accomplishments that we're proud of

  • Fetching over 1000 videos
  • Using spacy to find relevant tags
  • Incorporating the front and back ends for a MVP

What we learned

  • Learning javascript and typescript with React
  • Natural language processing techniques
  • Web development and design

What's next for EduBytes

  • Include more channels and their videos, possibly with a more structured database
  • Personalised recommendations with community driven
  • Specific related links -- Wikipedia isn't the most reliable source
Share this project:

Updates