Sentiment Analysis of YouTube Comments for Understanding User Engagement in Promotional Videos
Prabin Kumar Baniya -Student, School of Computer Science and Engineering, Galgotias University, Greater Noida, India
Abhishek Bhagat -Student, School of Computer Science and Engineering, Galgotias University, Greater Noida, India
Abstract—
YouTube is one of the major players digital advertisements. Advertisers can advertise their content/product on the YouTube and can target wide range of audience. There are generally two major ways using which advertisers can advertise their product on the YouTube. First way is, using the Google AdSense. Second way, is directly reaching the content creators and paying them to introduce/advertise their product in their video. Since, the like and dislike ratio of any YouTube channel is a great way to understand how good the channel is and it helps to take decision of whether advertising the product in that YouTube channel is good or bad. But with the change of YouTube policies, now we can’t see the number of dislikes in any video. This can be a great problem for the advertisers who want to advertise their product on YouTube and aren’t aware of which content creator fits best for them. To cope with this problem, we can use sentiment analysis to figure out how many videos are receiving positive and negative response on the videos uploaded by the creator. This tool will help the advertisers to find out which YouTube Channel would be best for advertising their product. To perform the sentiment analysis, we can use Naive Bayes Multinomial classifier to perform the sentiment analysis. A dataset containing the positive and negative comments is trained and used to build a model using this algorithm. We can expose this ml model in the form of an API and can create a frontend application which consumes this API and show the result in graphical way. As a result of this project, we shall expect a web application that ask for the YouTube video URL and after entering the video URL, the system starts scraping the comments of that particular video. Once the comments are scraped, we can use our model for finding sentiment for each comment. Based on that, a graphical result shall be shown which consists of how many positive, negative comments are found on that particular video. YouTube is one of the great places to advertise a product from any creator. Whereas, the likes-dislikes ratio plays a vital role in finding the right creator and asking to advertise the product. The like-dislike ratio helps to understand how much user engagement will be received on any particular channel.