We extracted data of Article 370 and US elections using the youtube api.Then we created Tf-Idf vectors and Google Universal encoder vectors for the entire data. The google colab for extracting the data and generating its vectors is "".480 videos were collected for US elections and 534 videos were collected for Article 370.
We then used K Means Clustering to generate clusters of the data.We observed that Google Universal encoder vectors generated more precise clusters as compared to Tfidf vectors. The clustering algorithm gave the labels as 0,1 and 2 for the 3 clusters formed.The three clusters formed in Artcile 370 data displayed the three different opinions "For removal of Article 370","Against Removal of Article 370" and "Neutral" opinion. The clusters fromed in US elections data displayed three different opinions-:"Democrat","Republican" and "Neutral".All this work is done in the google colab named "Backend_Article370.ipynb" and "". A final dataframe was generated which contained the video id,video title,video url and the labels for all the videos.These dataframes were then converted to csv files.This csv file was further converted to a Javascript dictionary.After that both these dictionary were imported in our final extension directory.
Clone the main directory and run the script.js in live server using the visual studio code.