Using data and text mining to find manipulated posts

Abstract

The steady growth of social media presence in society has increased the speed of information propagation across the globe. Because of wide information reach, social media platforms have become a prime place for information spreading through manipulative algorithms and bots. These activities have existed since creation of the internet, in forms of spam emails and scams, and now they have evolved. This dissertation focuses on how to extract large amounts of data from a popular social media website – YouTube, discusses methods for analysis of large volumes of data through popular text mining models, and proposes a system for information retrieval for detection of manipulation.