An Ensemble Approach For Fake News Detection
Abstract
"Fake News" is a term used to represent false news or propaganda-driven news which leads to the spread of misinformation via traditional media like print or non-traditional media like social media. This has become a major challenge in recent years due to the lack of computational tools. In this paper, we have proposed an Ensemble Approach for detecting fake news. Different classifiers are combined in order to filter the noise and to overcome the drawbacks of a single classifier. Extensive experiments are conducted on the dataset and different deep learning approaches such as Bag of Words, Sentimental Analysis, and Long-Short Memory Networks (LSTM) have been incorporated for the implicit and explicit features of text and ensembled using Max Voting by the use of weights. We have compared our results with existing models and our model gives accuracy of 92% on “Getting Real about Fake News” Dataset.