Polarity Trend Analysis of Public Sentiment on YouTube

Abstract

For the past several years YouTube has been by far the largest user-driven online video provider. While many of these videos contain a significant number of user comments, little work has been done to date in extracting trends from these comments because of their low information consistency and quality. In this paper we perform sentiment analysis of the YouTube comments related to popular topics using machine learning techniques. We demonstrate that an analysis of the sentiments to identify their trends, seasonality and forecasts can provide a clear picture of the influence of real-world events on user sentiments.