Using machine learning in social media systems

Abstract

Nowadays, vast use of Internet and especially, social media, as primary sources of information on everything is happening around the world, has unfortunately facilitated the spreading of fake news at the same time. Thus, everyone can alter real news and publish it on a news website or their social media account, or even invent news and promote it as real, misinforming and even disorienting in this way the public. For this reason, it is crucial to find ways to detect fake news as fast as possible, since fake news dissemination can sometimes be proved destructive, mainly as far as political and social issues are concerned, which have the stronger impact on people’s lives. Classification algorithms use is one way researchers have found in order to deal with this serious problem. In this thesis, we are going to present such a solution, which deploys Data Science and Machine Learning, in order to build a classifier for fake news detection. More specifically, after studying various articles concerning fake news classification, we are going to implement and evaluate our own classifier in a kernel created in Kaggle platform.