Geographical data from smartphones and Twitter messages could be used as a tool to assess the size of a large crowd of people, a study suggests.
Researchers at the University of Warwick said they used levels of smartphone activity along with geo-tagged tweets to successfully capture a fairly accurate crowd count in two locations in Milan, Italy - at an airport and a football stadium.
Phone data increased and decreased in close step with the number of people in each location, the researchers report in the journal Royal Society Open Science.
They called the results "a very good starting point" and suggested estimates of greater accuracy would be possible in the future.
"These are the numbers - the calibration examples - that we can draw on," says study co-author Tobias Preis. "Obviously it would be even better if there were training examples in other countries, other environments, other time periods."
That's because human behavior is not the same everywhere on the globe, he noted.
"But it's a very, very good base to build on, to provide initial estimates," Preis says.
Potential uses for the technique could include estimates of how many people were involved in a protest or rally, the researchers suggested.
The use of smartphone data could offer advantages over traditional methods of estimating the size of large crowds, says Warwick doctoral student Federico Botta, who led the study analysis.
"This is very quick," he says. "It does not rely on human judgment, it only relies on having the data related to mobile phones, or Twitter activity."
At the stadium, which is home to soccer teams AC Milan and Inter Milan, they compared the number of people based on ticket sales with three measures of smartphone activity: the number of calls and text messages, the level of Internet use, and the rate and number of tweets.
They did the same at the airport, matching phone data to the expected size of the crowd suggested by flight schedules.
Their margin of error was about 13 percent, they reported, which is considered a success compared with traditional estimations based in images, grids and human judgments.
The did acknowledge some limitation to the data the gathered, noting that only a subset of any population will be smartphone users, and not all areas where crowds may gather will have good cell coverage.
