Relationship Between Video Content and Audience Engagement: Evidence from BiliBili Power Uploader 2024
(A2026-133189)

AUTHORS

WEI JIE LAI, University of Macau; Lancy Mac, University of Macau

ABSTRACT

Focusing on Bilibili Power Uploaders (PUs) 2024, this study investigates how video title sentiment, content type, and recommendation status influence user engagement—measured by views, likes, coins, favorites, and total engagement—within Bilibili’s youth-driven ecosystem. Using a dataset of uploader and video metrics, this study applies negative binomial regression to analyze count-based outcomes and statistical tests for total engagement, controlling for follower count, video count, video age, and posting time. Results indicate that recommendation status significantly boosts all metrics, while positive title sentiment unexpectedly reduces views and likes but increases coins and favorites. Content types like Games and Music outperform others, highlighting the roles of algorithmic visibility and content diversity. These findings offer theoretical insights into platform-specific engagement dynamics and practical strategies for creators on Bilibili.