A Framework to Detect Twitter Platform Manipulation and Computational Propaganda

Manipulating public opinion using social media has become a pressing issue in this decade. Evidence of using organized social media manipulation campaigns have been taken place in 48 countries in 2018 alone and in each country, at least one government agency or a political party has been involved [1]. This is a massive threat to democracy and there's nothing new in governments carrying out propaganda, but using toxic messaging on a global scale and using new tools for amplifying while leveraging psychological methodologies is new. Facebook is being used heavily for such activities and other platforms like Twitter has been marked as a good option to spread misinformation and spam. The respective researchers say that some of these scenarios were taken place in countries which are new to social media and have been experimenting with computational propaganda and information control. It is impossible to rely on users to report such content and flag them as misleading, they would help the spreading of such information. This paper proposes novel methodology to detect such scenarios using machine learning and natural language processing techniques by predicting the credibility of the user profile and the credibility of the content. It uses existing research as foundation and builds new perspectives to look at the problem.