Bengali Fake News Detection Using Machine Learning

Abstract

This project based on NLP (Natural Languages Processing) techniques. The aim of this project is to identify the fake news of non-reputed news portal and to deliver the natural news or actual news to the news reader. As fake news misleads a great mishap among people so as a request of public demand we are going to run a model to detect all kinds of fake news. We collect data from reputed and non-reputed online news portal. This model build base on Bag of word (convert text format into vactorize format), tfidf matrix (extract facture) and RandomForestClassifier (for train and test dataset).By using this model we also create a predication where we test out site data is this data news fake or real. By using word tokenize we collect maximum number of keywords into our dataset for fake news and real news actually our model give result by comparing the dataset keywords. Our model achieves 86% accuracy.