Hongkai Mao

Sentiment Shift: Interplay between users and the community on a Chinese Movie-rating Website

Researchers have been increasingly interested in group polarization in online spaces recently. On many occasions, polarized sentiment is an important part of group polarization, and being negative is more common in groups(Zollo et al., 2015, Del Vicario et al., 2016). However, negative sentiment could be detrimental to the online community (Cheng, Danescu-Niculescu-Mizil, & Leskovec 2014, May). This work aims to study the sentiment drift in an online community and how users contribute to the sentiment drift differently in the hope of improving our understanding of emotion dynamics in a community. Ultimately, this work could serve the greater purpose of helping us prepare for proposing solutions to control potential negative emotion contagions. Using over 4 million movies from a Chinese Movie-rating Website, Douban movie, this study unveils the sentiment drift in that community and the interplay between users and the community. Our analysis reveals a sentiment shift on Douban Movie. Specifically, active users tend to contribute more to the community’s negativity. On average, users shift toward the negativity fast at the beginning period in the community and then slow down later, suggesting a fitting-in process. Our analysis also highlights new users’ influence on the community’s positive state. While previous research primarily focuses on misinformation accounts of online group polarization, findings of this work together show that polarization could rise without misinformation, suggesting the possible mechanism of group identity on group polarization on social media platforms.

Bio: I am a graduate student in the Computational Social Science program at The University of Chicago. My research interests include human behaviors on social media platforms, Human AI interaction, and social inequality. I actively combine tools like language models, deep graphs, and reinforcement learning with social science topics.