Majority Voting Technique for Fake News Detection

Abstract

A considerable number of people get their news from electronic sources. As there was a great increase in the usage of these platforms, more and more people are consuming the news from social media sources along with other websites. Thus, a variety of sites and sources has grown exponentially, allowing for easy and fast distribution of false information. Such deliberately generated lies fixing to deceive both the person and society are referred to as fake news Since the media plays an essential role in presenting fake information that alters public opinion and makes members of society take responsibility for unsupported facts. Currently the widespread of social media has worsened the level of fake news dissemination. To assess our methodology, we used popular machine learning classifiers such as Support Vector Machine (SVM),  Random forest (RF), Logistic Regression(LR), Naive Bayes(NB), Gradient Boosting, AdaBoost, K-nearest neighbor (KNN), Decision tree (DT), and    Extreme Gradient Boosting (XGBoost). We constructed a multi-model false news detection system using the Majority Voting approach, using the previously discussed classifiers to provide more accurate findings. Our strategy achieved an accuracy, of 0.9977%, recall of 0.998%, precision of 0.997%, and 0.997% of F1-measure, with a loss of 0.0022 according to the trial data. The assessment reveals that, in comparison to individual learning strategies, the Majority Voting approach produced more outcomes that were deemed acceptable.