Identification of Misinformation in News in Social Media

Abstract

Nowadays, social media platforms have become central to disseminating information, shaping public opinion, and influencing behavior in various areas. Sharing information and media with a large number of users is most effectively done through this method. However, it also has a major downside of sharing misleading and fake information, intentionally or unintentionally, that can harm others. The reason behind this is the absence of a proper way to identify whether a news story is authentic or misleading. Therefore, spreading misleading information on these platforms poses a significant challenge and requires a comprehensive understanding of identifying and controlling it. Researchers have attempted to identify fake news through various methods. The author’s approach involves distinguishing between disinformation and misinformation, with disinformation being more dangerous as it targets real individuals or entities and can potentially harm their reputations. The proposed solution to this problem involves using logistic regression and random forest classifiers to accurately determine whether the content of news is disinformation or misinformation. At the initial implementation phase, the accuracy of reliability detection in news was found to be 89%. However, identifying disinformation and misinformation has proved to have low accuracy due to the variability of the data set. Therefore, further development of the project requires proper optimization of these models.