Abstract

Since the outbreak of COVID-19, the pandemic has changed the lives of many people and brought dramatic motional experiences. Among many social media platforms, YouTube saw the most significant growth of any social media app among American users during the pandemic, according to the Pew Research Center on 7th April 2021. Exposure to COVID-19 related news can have a significant impact on user engagement on social networks. Different news may trigger different emotions (i.e., anger, anticipation, disgust, fear, joy, sadness, surprise, or trust), and a user may engage differently in response to the news. On YouTube, user engagement is manifested through actions such as liking, disliking, commenting, or sharing videos. During the pandemic, many users provide constructive comments that are encouraging, respectful, and informative to support each other. We applied sentiment analysis in the study to investigate different emotions and applied semantic analysis to investigate positive appraisal (i.e., encouraging, respectful, and informative) to identify salient factors that can motivate user engagement. The findings of the work shed light on how social network platforms could encourage constructive comments to help people provide emotional support to each other during pandemics through using positive appraisal in online news comments. The first research objective is to study the impact of sentiment valence of different emotions on people’s liking of news comments. News about COVID-19 on social networks may provide valuable information but also bring about public panic. In response to this COVID-19 related news, reviewers expressed their feelings by clicking the like, dislike buttons to the video and comments, or writing some comments under the video on YouTube. Some positive news was followed by comments expressing their anticipation, joy, and trust, while negative news might trigger sadness, fear, disgust, or anger. Our research focuses on sentiment analysis of news titles and the comments following each video. News title provides important information about the video, showing the summary of the video and allowing people to get a first glimpse of the content of the video. Through sentiment analysis of title and comments, correlations could be found between title/comments sentiment and user engagement. The second research objective is to investigate the impact of comments’ positive appraisal (i.e., encouraging, respectful, and informative content) on user engagement. The informative comments under the negative news have significant implications for the audience. They can be considered as a complement or judgment of the video content. Encouraging and respectful comments also help people build good conversations online. Our research focuses on semantic analysis of news titles and comments based on the three dimensions of positive appraisal and analyzes their impacts on user engagement to like the corresponding comment. We discuss the correlation between video title sentiment and the positive appraisal followed in the comments of the video to provide good conversations on the platform. A group of 38,085 online comments was collected from more than 400 different publishers from January 1st to January 30th, 2021, on YouTube. The dataset contains the most-viewed videos that were related to at least one of the following search queries: coronavirus, COVID-19, pandemic, or vaccine. NRC lexicon is adopted in the sentiment analysis to identify different emotions in titles and comments of the video. We adopt the topic modeling method and build a classifier from the Yahoo News Annotated Comments Corpus to identify constructive online comments for specific topics. We also measure inter-annotator agreements and compare the reliability of manual annotation and the classifier. We find that longer titles and sad emotions can obtain more likes on the comments of COVID-19 related news. During the pandemic, people tend to show their support when they find others are quite sad. We also expect to see correlations between some positive appraisals and user engagement.

Share

COinS