Content Analysis of Chinese Netizens' Public Opinion Based on Internet Comments: Case study of 160,000+ comments from Guan Video - a social political channel followed by four million people

Abstract

When a social event is reported, public opinions that an individual could access are predominantly limited by mainstream newspapers, social media, and how they portray the voice of their audiences. Recommendation algorithms from those platforms could be biased based on those channels' political standpoints. Therefore, the comments under news sections could be manipulated and not serve as an optimal proxy for reflecting public opinions. Social groups with different interests and standpoints are constantly conflicting with others. The measurement of social opinion can help us study the social demands behind public opinion and the social conflicts that come with the demands. Thus, content analysis of comments on Internet news articles proves effective in quantifying and understanding the different demands and conflicts in society more accurately, making it possible to govern a more developed and complex society. Carrying the goal to visualize abstract and subjective social opinions into measurable data, and to prove the effectiveness of applying data in assisting social studies, this research analyzes 160,000+ internet comments from Guan Video, a social-political channel that is followed by four million people through the Chinese video platform Bilibili, which has over 223 million Chinese users. With machine learnings IF-IDF model and Word2Vec model, this research proves the effectiveness of applying innovative quantitative measures in social research through the case study of netizens' opinions on education topics. Furthermore, we hope this research inspires further quantitative studies in humanity and society with the assistance of Natural Language Processing and other technologies.