Detection of Fake News on Twitter Using the Naive Bayes Model: A Brief Tutorial

The Internet has provided easy access, and the exponential growth of the information available on social media networks has made it intricate to distinguish between false and authentic information. The credibility of social media networks is also at stake, where the spreading of fake information is prevalent. Thus, it has become a research challenge to automatically check the information content and publisher to categorize it as false or true. Machine learning has played a vital role in the classification of information. This approach uses an API for collecting data from Twitter, state of the art as a reference to select features to separate fake and real users and tweets, and creating a labeled dataset. A Naive Bayes machine learning model is proposed to identify whether the social network Twitter news about Ecuador is real or fake during the “Paro Nacional” event from 12th June until 30th June 2022. Using a feature extraction and selection process to classify user accounts and tweets datasets as real or fake, to make them automated by training a Naive Bayes model. This work presents an implementation protocol that allows the development of a fake news detector system capable of differentiating between official and unofficial news of the “Paro Nacional” event. The results showed that the proposed methodology and model can identify with 62% accuracy real or fake accounts and with 49% accuracy real and fake tweets.