ABSTRACT

Fake news is a concerning problem in society presently and social media platforms are a primary source for its dissemination. In this paper, we set our sights on detecting fake news on social media using machine learning algorithms. Specifically, we analyse five popular classification algorithms (Naive B, SVM, Logistic Regression, K-Nearest Neighbors, and LSTM) on a fake news dataset from Kaggle. Our objective is to determine the optimal model for categorizing news articles as accurate or false depending on their substance. Including title, author, and article data. After testing the algorithms on the dataset, we found that LSTM was the optimal algorithm for this classification, with Logistic Regression coming in a close second. Moreover, we have added a new component to our research where we store the title, author, and article in a SQL database and display it as a blacklist. This enables us to easily identify and filter out fake news on social media.