Can the combination of deep learning (DL) – a branch of artificial intelligence – with social network analysis (SNA) make social media posts on extreme weather events a useful tool for crisis managers, first responders and government scientists? An interdisciplinary team of McGill researchers has brought these tools to the fore to understand and manage extreme weather events.
The researchers found that using a noise reduction mechanism, valuable information can be filtered out of social media in order to better assess trouble spots and assess user responses to extreme weather events. The results of the study will be published in the Journal of Contingencies and Crisis Management.
Immerse yourself in a sea of information
“We reduced the noise by finding out who was being heard and what sources were authoritative,” explains Renee Sieber, Associate Professor at the McGill Department of Geography and lead author of this study. “This skill is important because it is quite difficult to gauge the validity of the information shared by Twitter users.”
The team based their study on Twitter data from the March 2019 Nebraska floods in the United States, which caused over $ 1 billion in damage and widespread evacuations of residents. A total of over 1,200 tweets were analyzed and classified.
“By analyzing social networks, it is possible to determine where people get their information from during an extreme weather event. Through deep learning, we can better understand the content of this information by dividing thousands of tweets into fixed categories, such as: B. “Damage to infrastructure and utilities” or “Sympathy and emotional support,” says Sieber. The researchers then introduced a two-tier DL classification model – a first in terms of integrating these methods in a way that could be useful for crisis managers.
The study highlighted some issues related to using social media analytics for this purpose, notably the failure to notice that events from tagged records like CrisisNLP are far more contextual than expected, and the lack of a universal language to categorize them of terms to which this relates crisis management.
The preliminary investigation conducted by the researchers also found that a celebrity appeal played a prominent role – it was indeed the case in the 2019 Nebraska Floods, which saw a tweet by pop singer Justin Timberlake shared by a large number of users Although this was not the case, it is proving useful for crisis managers.
“Our results show that the information content varies between different types of events, contrary to the assumption that there is one universal language for categorizing crisis management. This limits the use of labeled records for only a few types of events, as the search terms can change from one event to another. “
“The large amount of social media data that the public contributes to the weather suggests that it can provide vital information in crises such as blizzards, floods, and ice storms. We are currently investigating the transfer of this model to different types of weather crises and correcting the shortcomings of existing monitored approaches by combining them with other methods, ”says Sieber.
About this study
“Using Deep Learning and Social Analysis to Understand and Manage Extreme Floods” by Renee Sieber and others. was published in the Journal of Contingencies and Crisis Management.
This study was funded by Environment Canada.
About McGill University
Founded in 1821 in Montreal, Quebec, McGill University is Canada’s top ranked medical doctoral university. McGill is consistently rated as one of the best universities in the country and abroad. It is a world-renowned university with research activities in two locations, 11 faculties, 13 vocational schools, 300 study programs and over 40,000 students, including more than 10,200 doctoral students. McGill attracts students from over 150 countries around the world. 12,800 international students make up 31% of the students. Over half of McGill students speak a mother tongue other than English, including about 19% of our students who say French is their first language.