Machine Learning Techniques for Identifying Textual Propaganda on Social Media: Development of a Detecting Digital Manipulation System

Abstract

The dissemination of propaganda on social media presents a significant challenge in today’s digital age. Utilizing advanced tools and diverse methods, propaganda aims to influence public opinion on a massive scale. Social media platforms serve as prime channels for such messages, leveraging sophisticated strategies to shape public perceptions and attitudes. This research aims to develop an advanced system capable of evaluating whether the content disseminated on these platforms qualifies as propaganda. The hypothesis suggests that it is possible to distinguish propaganda from non-propaganda texts on social media by analyzing specific linguistic features. Employing advanced linguistic analysis and machine learning methods, this detection system achieves approximately 70% accuracy, indicating its promising potential for effectively identifying propaganda. This approach could significantly enhance the transparency and reliability of online information, encouraging a more informed and critical use of social media. 

Keywords: Propaganda Detection, Social media, Machine Learning Techniques, Linguistic Analysis, Digital manipulation,