Sentiment Analysis on YouTube Data: A Comparison of TextBlob and VADER

This article tried to compare the performance of TextBlob and VADER on analyzing comments under YouTube videos and measured the correlation between using sentimental analyzers to predict the quality of videos and using the like/views ratio to compare the quality of videos. People are becoming busy, and they rely more on watching videos to have fun, acquire new knowledge, and exchange thoughts. Finding a way to estimate the quality of the videos becomes important. 18000 comments are collected from 60 videos from 6 categories. TextBlob and VADER are used to analyze these comments. Scores are given by these two sentimental analyzers. The scores of each video are compared to the likes/views ratio to predict the quality of the videos. Charts are made to analyze the trends. The result is that TextBlob is more suitable for this model because the scores it gives are more stable and have a positive correlation to the likes/views model.