Hack2Educate-GenZ

Watch the demo :

Team GenZ

The Team Members of GenZ are:

• Harsh Raj
• Raj Vardhan Tomar
• Arpita Kesharwani
• Rahul Kumar Mishra

We are working on an AI/ML theme.

• We will be given a Youtube video and then our tasks are following :-
  • Determine if it is an educational video or not.
  • Determine which category or subcategory of BE it would belong to.
• Accuracy: 90%+ on a custom dataset that will be provided after proof of concept.
• Should be able to run in near-real-time on a server.

Problem solving approach:-

• Used Natural Language Programming to solve this problem

• Project 1

  • To learn how to use Application Programming Interface (API) to extract some data of YouTube channels and videos.
  • To extract data from Educational, Entertainment, Technology and Motivational channels.

• Creation of Dataset

  • Data was extracted from approximately 1.5 Lakh YouTube videos from 45 different channels.
  • Separation of educational and non-educational keyword was done so that it can be checked if the video is educational or not.

• Algorithm

  1. To classifying the video as educational or non-educational.
     • We extracted the title and description of a YouTube video using its link.
     • Removed the punctuations, links and emojis from the extracted data.
     • Created a dictionary of keywords by splitting the data word by word.
     • Removed all the non-educational keywords from the video's keywords.
     • Compared the remaining keywords from our educational keyword dataset and using the algorithm it is calculated that what is the percentage of the video to be educational.
     • And if the percentage calculated is above certain number then the video is declared to be educational.
  2. To predict from which category or subcategory of Beyond Exams the video belongs.
     • Defined a list of keywords for each of the subcategory
     • Used the same procedure as first algorithm to check which subcategory of BE the video may belongs to.

Final demo

• The final outcome of the project will be a website where link of any YouTube video can be pasted and a result will be displayed on the page to show if the video is educational or not.

PREREQUISITES

• Python 3.x
• pip3
• flask

CLONE

git clone https://github.com/rajharsh18/Hack2Educate-GenZ

RUNNING

cd Hack2Educate-GenZ

cd TBE-Website

pip install -r requirements.txt

Create a YouTube v3 API key
Paste it in the app.py file

Run the app.py file

The webpage will be opened on your localserver