Tackling Misinformation Through Tweets: A Comparative Study of Various Machine Learning Approaches

Fake news or deceptive journalism is one of the most researched and worked-on projects in the machine learning world. In our modern and expeditious world, getting information and knowledge is easier to access on the internet than other sources. Social media, websites, and blogs have completely replaced newspapers and articles for the majority of the population. This in fact has caused the rise of more destructive issues. Newspaper agencies always have a verified and trustworthy news source, but with the rise of internet usage and easy access, fake news is at an all-time high. Numerous studies have been conducted that document the effect of misleading and inaccuracy information on the general public and how it has affected the daily commute and working of the common people. Fake news is basically fabricated posts with fabricated evidence with visual content that is completely edited using the best software available. This study employs a number of machine learning approaches to address this issue. The research was conducted on the Fake and Real News dataset, tackling misinformation through tweets. Several machine learning algorithms like Logistic Regression, Gaussian Naive Bayes, Passive-Aggressive Classifier, and LSTM machine learning models were used to tackle this challenge. The LSTM model, with an accuracy of 99.23% came out as the best model for segregating real news from fake news.