The constant stream of tweets available on Twitter can make it seem like an overwhelming and chaotic tool. An international team of researchers, however, hopes to form order out of that chaos, using a newly developed algorithm designed to read in Twitter data, identify disaster-related events, and generate real-time summaries to assist in disaster response activities.
Featuring a partnership between researchers at Penn State, the Indian Institute of Technology Kharagpur and the Qatar Computing Research Institute, the algorithm seeks to take advantage of the rapid updates available. Using a set of 2.5 million tweets made in the aftermath of three major disasters – the 2014 Typhoon Hagupit, the 2014 flood in Pakistan, and the 2015 Nepal earthquake – a machine learning system was trained to identify smaller events based on word pairings for different smaller disaster events, like “person trapped” and “bridge collapse”. After training was completed, this tool was then reapplied to rapidly updating data to identify the occurrence and frequency of these terms to help extract information to pass along to first responders and other associated planners.
“The best source to get timely information during a disaster is social media, particularly microblogs like Twitter,” says associate dean for research at Penn State’s College of Information Sciences and Technology of the approach.
Sources:
https://i-hls.com/archives/84328
