Abstract: The rapid proliferation of social media platforms such as Twitter creates a fertile ground for misinformation and fake content. The detection of fake tweets is a highly complex challenge because of the short length, diverse topics, and linguistic nuances that these short messages carry. The present research introduces Article Authenticity Analyzer (AAA), a novel system, for effective identification and classification of fake tweets. The proposed analyzer integrates advanced NLP techniques, user behavior analytic, and social network analysis to provide a holistic detection framework. The system extracts contextual and semantic features from tweet content by leveraging transformer-based models like BERT, whereas user behavior analysis evaluates credibility based on metadata such as account age, posting frequency, and network interactions. Graph based techniques are used to uncover coordinated misinformation campaigns. The AAA achieves state-of the-art performance with an accuracy of 92% and demonstrates robustness across multiple datasets. This paper discusses the methodology, experimental setup, and real-world implications of deploying the AAA in combating fake news on social media platforms.
|
DOI:
10.17148/IJARCCE.2024.131255
