Sina Weibo is China's largest microblogging platform, launched in 2009. With over 300 million users (8× Twitter's size as of 2012), Weibo has distinct structural and cultural characteristics that shape misinformation propagation differently than Western platforms. The platform has an official rumor-busting service, enabling high-quality data labeling for verification research.
Platform characteristics and implications for misinformation
Scale and reach: Weibo generates over 100 million microblogs per day, making it a critical venue for tracking Chinese-language information spread.
Structural differences from Twitter:
- Hybrid design: Weibo combines features of Twitter (microblogging) and Facebook (social network), with support for images, videos, and embedded media.
- Retweet behavior: Trends on Weibo often feature media content (jokes, images, videos) rather than news and current events as on Twitter.
- Official verification: Weibo operates an official rumor-busting service account, confirmed false rumors are publicly marked, enabling authoritative ground-truth labeling.
Linguistic and cultural factors:
- Chinese language features (no case sensitivity, different punctuation norms) differ from English-language misinformation signals studied on Twitter.
- Domestic focus: The majority of rumors tracked concern China-related events; geographic scope (domestic vs. foreign) is a strong predictor of rumor veracity.
Key papers in this wiki
- Yang et al. (2012) — Automatic Detection of Rumor on Sina Weibo: First peer-reviewed study of rumor detection on Sina Weibo; proposes client-program and event-location features; achieves 78% accuracy with platform-specific feature engineering.
- Ma et al. (2016) — Detecting Rumors from Microblogs with Recurrent Neural Networks: Evaluates RNN-based rumor detection on both Twitter and Weibo; GRU model achieves 91% accuracy on Weibo (4,664 events) with early detection capability within 12 hours.
