Classification of Fake News Using Multi-Layer Perceptron

"Fake News (FNs) is defined as a made-up story to deceive or to mislead." The problem of FNs spread widely in recent years, especially on social media such as Facebook, Twitter, and other sources like webs and Hogs. It has become a significant problem in society as a result of changing people's ideas and opinions about the direction of this news. In this paper, FNs detection can be proposed by using the Term Frequency-Inverse Document Frequency (TF-IDF) as features extraction, and Multi-Layer perceptron (MLR) algorithm as a classifier. Two phases (feed-forward and hack-propagation) are used with a three-layers, which are (input layer, one hidden layer, and output layer). After running our proposed algorithm on a FNs dataset, the classification accuracy achieved equals 95.47%.