YouTube Video Classification based on Title and Description Text

Abstract

YouTube has a library of millions if not billions of videos and keeping a track of the types of videos for effective retrieval and use can be quite difficult. YouTube videos can be classified into different classes based on the title and descriptions of the videos. To classify so many videos, an effective scalable algorithm is required. This can be achieved by using a Random Forest Classifier along with Natural Language Processing techniques like Bag of Words, Word Stemming etc. This paper also discusses method to scrape YouTube videos using packages like selenium, requests and Beautiful Soup for videos and their metadata. At the end we discuss various evaluation metrics for Random Forest Classifiers.