Nowadays, the use of social networking sites is becoming more popular especially in the disaster like situations. Social media has changed the way of communication. People now get news through it. Apart from all these positive features, social media is also slowly turning out to be a live-saving tool. Various social media platforms, including Twitter, blogs, News aggregators, etc., include heterogeneous material in a variety of forms. An oversized quantity of helpful data is shared on social networking throughout associate emergency, together with the users’ sympathies and opinions. So, normally people uses their natural language in which they speak for communication which computer doesn’t understand. Therefore, there should have a reliable methodology which required for extracting helpful information (Context) from the natural language that people normally used for the communication and sharing their available information regarding current scenario happening nearby them during the disaster. Further down this information can categorized into situational and non-situational information. Situational information indicates a current consequences happening due to disaster and where immediate action is required which may reduce further losses that may likely to be happened. Social media today are crucial for disseminating information from the real world, experiences from daily life, and ideas through online groups and networks. Real-time events like disasters, power outages, traffic, etc. can leverage this information. Due of the noisy data, unrelated data, and data in many formats present on social media, analysing and comprehending such information can be difficult. Because of this, this study examines and classifies numerous event detection techniques in various forms of social media.

Disasters, RoBERTa model, situational information, Event detection