Need key info in a disaster? There isn't an app for that, U of C researchers say
In this May 7, 2016 file photo, a wildfire burns south of Fort McMurray, Alberta. (Jonathan Hayward /The Canadian Press via AP, File)
Lauren Krugel, The Canadian Press
Published Monday, June 19, 2017 2:42PM EDT
Last Updated Monday, June 19, 2017 5:58PM EDT
CALGARY -- A study that analyzed nearly 70,000 tweets sent in the thick of the Fort McMurray wildfire shows smartphone emergency apps don't provide much of the information people need during a disaster.
Maleknaz Nayebi, a doctoral candidate at the University of Calgary's Schulich School of Engineering, recalls going online in May 2016 to figure out how she could help more than 80,000 people fleeing the flames in northern Alberta. She realized many evacuees were flocking to social media for help.
"As local radio stations went off the air and websites failed, social media became the crisis' unofficial emergency broadcast system," researchers wrote.
The findings were published through the International Conference on Software Engineering, which was held in Buenos Aires last month.
Communications staff for the Regional Municipality of Wood Buffalo's Twitter account were praised for putting out a steady flow of information throughout the disaster.
But Nayebi, the study's lead author, said answers to specific questions were easily missed in the deluge of Twitter traffic.
Apps made specifically for wildfires and other emergencies have the potential to help on that score, but Nayebi and her fellow researchers found the ones on the market are lacking.
The researchers developed software called MAPFEAT -- an acronym for Mining App Features from Tweets -- to sift through 69,680 tweets sent between May 2 and May 7, 2016, and isolate some of the top concerns and questions evacuees had.
The most common queries were compared against features provided by 26 wildfire and emergency apps the researchers found in Google and Apple stores. Those included apps created by the Alberta government, local governments in Australia, the Red Cross and the International Association of Fire Fighters.
There were followup studies with the general public about what app features would have been most helpful.
The researchers discovered a big mismatch. None of the top 10 concerns was addressed by the apps and only six of the top 40 were covered.
It was found some of the app features that would have helped the most had to do with fire alarm notifications, requests for food and water and where to find the nearest gas stations.
"We are developing software that people really don't need," said Nayebi.
She said she'd like to bring the MAPFEAT methodology into practice.
"We need to have more authorities on board. We would love to get in contact with some NGOs or organizations or city governments or governments of different countries."
Guenther Ruhe, a contributor to the research and Nayebi's PhD supervisor, said MAPFEAT can have uses beyond disasters. He suggested financial or health-care-related apps.
"It is a very innovative and unique technique to better understand the needs of customers."
It may seem straightforward, but Ruhe said knowing what users want can be the most tricky part of software development.
"I'm really excited about this and it's really an excellent example about how research can not only end up in publications and citations, but it can have an impact on society."