An Innovative and Implementable Approach for Online Fake News Detection Through Machine Learning

摘要

One should recollect the USA 2015 and 2016 U.S. presidential election cycle dealt with numerous scandals which were triggered by the forged news articles that blowout through the social media like Twitter and Facebook. When it was found that these articles were purposefully uploaded for financial and political gain, it's become evident that bogus news has to be identified and removed to prevent public from being deceived for someone's personal gain. This study builds a supervised machine language model to detect the fake news articles published during 2015 and 2016 U.S. election cycle. The data set contains identical number of bogusand factual news. The standard set of machine learning algorithms like K-Nearest Neighbors, Support Vector Machine, Naive Bayes and Passive Aggressive Classifier are trained using either the title or the content of the article. There results show that the PAC classifier produces the highest accuracy of 94.63% over the other three classifiers using diagram term frequency.