The project examines several large datasets political tweets in English, Russian, Korean, Farsi, German, Spanish, Bengali, Catalan, Arabic, etc. posted by accounts that have been identified by Twitter as participating in an astroturfing campaign, i.e. an hidden attempt to influence public opinion by pretending to be ordinary users. We will examine how the behavior of participants differs from that of regular users and accounts, and how the differing patterns can help us identify ongoing efforts to secretly influence online public opinion worldwide.
This project is an extension of "Political astroturfing in South Korea: examining how the secret service manipulated twitter during the 2012 presidential election." It is supported by a UROP grant, and students will therefore be able to suggest what additional data should be acquired to analyze the campaigns and their impact.
Most students will be in charge of analyzing one particular campaign or an aspect of it. This analysis will be guided and supported, and students will receive instructions and code to perform certain types of analyses at each step. They may be required, for instance, to read through some of the tweets posted by these astroturfing accounts and describe what arguments and strategies these accounts use. Do they copy-paste each other's comments? How do they interact with other regular accounts? How does their behavior change over time? Do different group of accounts pursue different strategies? The students can chose the campaign that suits their language skills.
Students will also acquire the necessary statistical or programming skills to get involved in the process of downloading and cleaning the data, and perform statistical analyses comparing the campaign participant's tweets with those of regular users.
1. Gain basic knowledge on how to analyze social media data
2. Learn how governments and parties influence public opinion using social media
3. Basic knowledge of how twitter works "under the hood"
4. Learn how to conduct simple statistical analysis and social network analysis on such data.