Fake News Detection Using Machine Learning Approach

Abstract

As internet usage rises, individuals are increasingly encountering false information. To gain fame and benefit financially, clickbait news sites and social media propagate misleading details to mislead the public. Engaging stories surrounding trending issues have emerged. While the proliferation of misinformation has intensified globally in recent times, various new methods for identifying and categorizing it have been developed. Not much research has been conducted on misleading information in English news reports. Bengali news was reported. This article examines forgeries in Bengali. The South Asian context is taken into account when categorizing news. Over5200 million individuals speak Bengalias their native language, and it defines their lifestyle. Our primary goal for this research was to begin an analysis of the assumptions betweenML and DL. The classifiers utilized in this instance included RandomyForest, SVM, DecisionbTree, XGB, Gradient BoostiClassifier, and Ada BoostyClassifier. The highest accuracy attained was589% by the GB. Subsequently, we employed several establisheddeep learning methods to carry out the second phase of ourbexperiments. Where we have utilizedvRNN,mLSTM, Bi-LSTM, GRU,mBERT. We have thoroughly presented the comparison of models to achieve the best evaluation outcome. A7comparative analysis was conducted to illustrate the comparison of other works. Which we believe were helpful in grasping the overall intent of our efforts. With the introduction ofLDeepuLearning techniques, our RNN-based model has attained a total accuracy of894%. Through which we present this article in the detailed exposition of our complete process.