Abstract:
People are getting more exposed to fake news as their use of the web increases. to get notoriety while profiting from clickbait news outlets and social media, The media promotes incorrect information to deceive people. fascinating content about a current topic Though the spread of incorrect information has recently become more serious throughout the world, several existing techniques for categorizing and detecting have recently been created. There hasn't been a lot of research done on misleading news in an English news report. There was coverage of Bengali news. In this article, we look at Bengali forgeries. When classifying news, the South Asian context is considered. More than 200 million people use Bengali as their first language, and it is their way of life. Communication necessitates are fundamental in Bengali. Our main intention towards this research was to initiate an interpretation between the ML and DL assumptions. The machine learning classifiers which were used in this case were Random Forest, SVM, Decision Tree, XGB, Gradient boost classifier and Ada boost classifier. The best accuracy was achieved by the GB which was 89%. And afterwards we have used various well known deep learning approaches to conduct our second stage of experiments. Where we have used RNN, LSTM, Bi-LSTM, GRU,BERT. Then we comprehensively shown the models comparison to product the best evaluation result. There was also a comparative analysis to show the otherward works comparison. Which we think were beneficial in case of understanding the whole purpose behind our work. As Deep Learning methods were initiated our model with the base of RNN has achieved overall 94% accuracy. By which we propose this article in the elaborate discussion of our full procedure.
