Search

International audience; Rapid location and classification of data posted onsocial networks during time-critical situations such as naturaldisasters, crowd movement and terrorism is very useful wayto gain situational awareness and to plan response efforts.Twitter as successful real time micro-blogging social media, isincreasingly used to improve resilience during extreme weatherevents/emergency management situations, including earthquake.It being used during crises by communicating potential risksand their impacts by informing agencies and officials. Thegeographical location information of such events are vital torescue people in danger, or need assistance. However, only fewmessages contains there native geographical coordinates (GPS).So identifying location is a real challenge with Twitter data duringcritical situations. Identification of Tweets and their preciselocation are still inaccurate. In this work, we propose to usesemi-supervised technique to utilize unlabeled data, which is oftenabundant at the onset of a crisis event, along with fewer labeleddata. Specifically, we adopt an iterative Random Forest fittingpredictionframework to learn the semi-supervised model.