Using geosocial applications to monitor accident occurrences at the country’s
busiest intersections, a team of researchers found that police response is
sometimes slow in these areas.
Led by Ben-Gurion University of the Negev
doctoral student Michael Fire, the team compiled their findings into a report
titled “Data Mining Opportunities in Geosocial Networks for Improving Road
Safety,” which they originally presented at the 27th Convention of Electrical
and Electronics Engineers, held in Israel this November.
Working with
Fire was scientific developer Dima Kagan, Dr. Rami Puzis, Prof. Lior Rokach and
Prof. Yuval Elovici – all from the university’s Telekom Innovation Laboratories
and Information Systems Engineering Department.
Using geosocial
applications available on smartphones – such as Waze, FourSquare, Facebook
Places, Google Latitude and Mobli – can be an inexpensive yet crucial tool in
compiling anonymous information about transportation, the report argued. In
particular, the researchers harnessed the capabilities of the application Waze
to look at user reports in areas inundated with traffic accidents, to examine
whether there tended to be a significant police presence at these sites.
Established by a team of Israelis, Waze is a traffic and turn-byturn navigation
application for smartphones, which allows drivers to share realtime traffic
updates, accident updates and road changes with each other.
“We show how
information created by the Waze’s user community helps to identify dangerous
intersections and locations that are plagued on a daily basis by reoccurring
accidents,” the report’s introduction said.
The researchers specifically
looked at the accident reports, traffic data and “police nearby” reports that
users compiled at these dangerous intersections over 31 days in the months of
May and June 2012, according to the paper.
Because Waze has more than 1.1
million active users in Israel, the researchers said they felt that the
application might provide a good level of coverage for Israel’s traffic
hubs.
In total, the researchers collected 5,369 accident reports and
29,789 “police nearby” reports, dividing them into a grid of cells based on
geographical area. For each area, they then providing ratings from zero to 31,
based on the number of days accidents occurred there, as well as corresponding
scores to show how many police nearby reports came in.
At the conclusion
of the process, the researchers found 2,743 areas with at least one accident
during the time period, 312 with a rating of 3 or more and 19 with a rating of 7
or more. Meanwhile, 579 locations had at least five recurring accidents during
the 31 days, and these same places were responsible for 5,156 of the accidents
reported – more than half of all the reported accidents, according to the
paper.
The researchers then pinpointed 3,555 locations where users had
reported “police nearby” statuses at least 15 times. After reviewing the data,
the team discovered that at least 75 percent of the 20 areas that had received
the highest scores for recurring accidents were intersections, the report said.
Likewise, 40% of the 20 areas that received the highest scores for police nearby
statuses also were intersections.
Using a linear regression model, the
researchers then determined the correlation between the number of accident
reports and the number of “police nearby” reports in each area unit. The models
revealed that approximately 67.9% of the accidents did not include police
intervention, and average police response time for an accident was 28.66
minutes.
“According to the data, police response time was sometimes
slow,” Fire said following the report. “There were also numerous instances where
the police were manning quieter intersections, while busier intersections went
unmonitored.”
Identifying the problematic areas could help improve police
deployment in the future as well as prevent accidents, the researchers
concluded.
Going forward, the team wrote, the geosocial applications
might be used for further research, such as measuring the influence of speed
cameras on accident rates.
“Our analysis could be used by the police to
see if they are manning the busiest and most dangerous intersections,” Fire
said.
|