Automated detection of fake news

Abstract

During the last decade, the social media has been regarded as a rich dominant source of information and news. Its unsupervised nature leads to the emergence and spread of fake news. Fake news detection has gained a great importance posing many challenges to the research community. One of the main challenges is the detection accuracy which is highly affected by the chosen and extracted features and the used classification algorithm. In this paper, we propose a context-based solution that relies on account features and random forest classifier to detect fake news. It achieves the precision of 99.8%. The system accuracy has been compared to other commonly used classifiers such as decision tree classifier, Gaussian Naïve Bayes and neural network which give precision of 98.4%, 92.6%, and 62.7% respectively. The experiments’ accuracy results show the possibility of distinguishing fake news and giving credibility scores for social media news with a relatively high performance.