Intelligent Fake News Detection Model using Machine Learning Techniques

Fake news includes wrong or inaccurate information that is published as real news. Detection of such news deals with identifying deliberate deception or hoax news propagated by public (print and broadcast) or digital media. It frequently aims to harm a person's or entity's reputation, and also contains bogus tales about a variety of topics that seem to be legitimate by driving attention to social media platforms. Several Machine Learning (ML) and Deep Learning (DL) models are employed in detection of fake news. The existing ML models rely on an unorganized database which does not support data augmentation. This reduces the accuracy of system. In this paper, the proposed scheme draws raw data from KDnuggets and ISOT datasets and categorizes them such that the data can be updated frequently. For every categorized dataset, Naive Bayes (NB), Support Vector Machine (SVM), Logistic Regression (LR), K-Nearest Neighbor (KNN), and Random Forest (RF) models are used for detecting fake news effectively. New data can be added to the database once it is validated by the administrator. User information can be in the form of text, photos or video (converted to text) and verified. The proposed ML-based model applied on the categorized dataset offers a testing accuracy of 99%.