Reproduce “Battling the Internet Water Army: Detection of Hidden Paid Posters” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Cheng Chen, Kui Wu, Venkatesh Srinivasan, Xudong Zhang
This study presents the first systematic analysis of the "Internet Water Army" phenomenon in China, where individuals are paid to manipulate online opinions. Researchers developed a detection system utilizing a unique combination of insider knowledge and machine learning, achieving 95.24% precision and 73.17% recall by analyzing both behavioral patterns and content similarity.
Key insights
- Human-driven opinion manipulation campaigns, like the "Internet Water Army," are highly organized and exhibit distinct behavioral patterns compared to normal users or automated bots.
- Non-semantic features such as reply percentage, average posting interval, and account active days provide indicative behavioral signals for identifying paid posters.
- Semantic analysis, particularly comment content similarity, is a highly discriminative feature for detecting paid posters due to their tendency to reuse or slightly modify content.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.