Feature-Based Supervised Classifier to Detect Rumor in Social Media

Abstract

Social media is the most important and powerful platform for sharing information, ideas, and news almost immediately. With this, it also attracted antisocial elements for spreading and distributing rumors that is unverified information. Malicious and intended misinformation spread on social media has a severe effect on societies, people and individuals, especially in case of real-life emergencies such as terror strikes, riots, earthquakes, floods, war, etc. Thus, to minimize the harmful impact of rumor on society, it will be better to detect it as early as possible. The objective of this research and analysis is to develop a modified rumor detection model targeted for the proliferation of any malicious rumors related to any significant events. It is achieved through a binomial supervised classifier. The classifier uses a combination of explicit and implicit features to detect rumors. Our enhanced model significantly achieved it with 85.68% accuracy.