Date of Award
12-2023
Document Type
Thesis
Publisher
Santa Clara : Santa Clara University, 2023.
Degree Name
Master of Science (MS)
Department
Computer Science and Engineering
First Advisor
Yuhong Liu
Abstract
Predicting the popularity of the YouTube video-sharing site in its early stages is crucial for journalists and researchers, as it helps them uncover disinformation. In addition, research on the number of comments in news and comments has always been an academic topic that has attracted much attention. These Youtube political commentary and News channels that use Chinese as the primary language have a strong influence on the first generation of Chinese immigrants in the United States. However, under our monitoring, we found that it is difficult for these videos to appear in the viral video list defined by YouTube API. In this paper, we present our own metrics based on comments of virality and early detection model focused on Chinese political commentary and News Youtube channels. And we collected data from these related channels for analysis, and propose our own metric to measure and detect. We achieve 0.81 F1 Score as early as the third hour of the Youtube video publish. We demonstrated that the number of comments per hour is a powerful feature independent of any models.
Recommended Citation
Zeng, Yulin, "Early Detection of Virality in Chinese-Language on YouTube focus on comments" (2023). Computer Science and Engineering Master's Theses. 37.
https://scholarcommons.scu.edu/cseng_mstr/37