Abstract:

Depending on the developing technological possibilities, there have been significant changes in the field of communication. With the widespread use of the Internet, traditional means of communication have been replaced by new technological methods that enable quick and easy access to information. At the forefront of these technological innovations is social media platforms that allow users to interact interactively. Among the many social networks offered to users, Twitter is preferred by both users who want to follow the agenda and news sources who want to deliver their news quickly to the target audience, as it is a suitable platform for written and visual journalism. This platform, which allows the news to spread and interact rapidly among people, has some disadvantages as well as advantages. Some of these are the difficulty of putting fake news into circulation and preventing them due to the inability to control the news. In this study, machine learning methods were used to detect fake news on Twitter. An exemplary topic was selected and fake and real news about it were identified. Decision trees and Naive Bayes methods were used in the study. The results of the study were compared according to the confusion matrix and F1 score method. The F1 score of the decision tree method was 0.829, and the Naive Bayes method was 0.883. According to these results, it has been seen that the Naive Bayes method is a more successful method for detecting fake news on Twitter. With this study, it will be possible to detect fake news on Twitter and take precautions.

Keywords: