Does it make you sad? A lexicon-based sentiment analysis on COVID-19 news tweets

Abstract

This research utilizes a lexicon-based sentiment analysis to reveal the emotions conveyed by various news media and analyze the differences among the media. NRC Affect Intensity Lexicon is utilized; it provides the emotions of words in 8 categories of emotions. The lexicon is modified to reflect the context of the news, i.e., COVID-19 pandemic. Tweets are collected from four Indonesia’s news media and three international English-language news media. Each tweet from a news media is assigned a total emotion score for each of the 8 emotions. The scores are averaged for each day to obtain the Daily Emotion. Based on the Daily Emotion, the Dominant Daily Emotion may be identified, i.e., the emotion with the highest average score in a particular date. The Dominant Daily Emotion visualization shows that the dominant emotions in Indonesian-language media are Sadness and Trust, while Fear and Trust are considerably more pronounced in English-language media. Furthermore, among Indonesian-language media, kompascom is significantly more intense than the others in Joy, Sadness, Fear, and Trust. Among English-language media, XHNews is more intense than others in conveying Fear, Sadness, and Trust; while XHNews and timesofindia are both the most intense in Joy. The sentiment analysis in this research concludes that the Emotion Mix of news media are significantly different. Furthermore, news media also convey significantly different levels of emotion intensity in reporting COVID-19 news.