Abstract:
Fake news detection is as important as cleaning crime from the society. In this digital era
spreading violence and manipulating people deception towards an issue become very easy
than ever. More and more authentic news are mutilated into fake news for some peoples
own mean .In our day to day life we use many online platform for both entertainment and
other uses .In those places we are continuously bombarded with fake news. This toxicity
needs to be stop for our own good. And to stop the spreading of fake news, detecting it is
the most inescapable part. Our research is based on that inescapable part. We tried to create
a model for identifying fake news with a very big dataset of Bangla news data. All the data
we have collected are from various online sources .Also we have used both machine and
deep learning methods for our study. In machine learning method LR, SVM and RF shows
95% of highest accuracy .On the other hand in deep learning method LSTM and BERT
give the best accuracy of 93%. Although we have used many methods but few methods we
couldn’t use for our low performance device. However an application can be develop with
this research to identify the news either fake or authentic.