/

Japan considers crime prediction system using big data and AI

JIJI, AFP-JIJI, Kyodo

The government and the police are discussing the idea of developing a computer system that can predict street crime by utilizing big data and artificial intelligence.

They hope such a system will be able to show them where and how to take greater measures to prevent crime.

Street crime prediction “has already achieved results in Europe and the United States,” said Mami Kajita, who established the data-analysis company Singular Perturbations Inc. last year in hopes of developing a Japanese version of the methods used in the United States.

In some parts of America, the police have ramped up patrols in areas where AI-based systems predicted crime was more likely to happen, achieving a reduction of 20 percent on average, Kajita said.

A more cautionary tale comes from China, where the government is racing ahead to use big data and facial recognition technology to surveil the population.

According to Human Rights Watch, an advanced system called the Integrated Joint Operations Platform is pooling data on people from many sources. It has already been deployed in the restive western Xinjiang region, where it is being used to flag and detain people deemed potential threats.

Kajita, 35, who studied theoretical physics at the University of Tokyo, once lived in Italy. Her husband, also a researcher, was eventually transferred there as well. They often fell prey to pickpockets.

She realized that people are vulnerable to crime when they are not familiar with their surroundings. She then came up with the idea of applying her mathematical research on natural phenomena to the realm of urban phenomena to prevent crime.

Kajita scrutinized the Tokyo Metropolitan Police Department’s email regarding cases of molestation, theft and suspicious people. Combining the data with her own original algorithm, Kajita predicted which areas were likely to see high levels of crime.

Despite her limited data, the predictions mostly matched areas where crimes actually occurred. Her accuracy is likely to improve if additional factors such as weather, traffic and Twitter posts are incorporated.

“We need to continue examining crime mechanisms unique to Japan,” Kajita said.

“I hope to create a smartphone app for citizens in order to reduce the number of crimes, even if only slightly,” Kajita said.

Kanagawa in January announced plans to become the first prefecture in the nation to introduce predictive policing, and hopes to put such a system in place on a trial basis before the 2020 Tokyo Olympics.

The AI-based system would employ a “deep learning” algorithm that allows the computer to teach itself by analyzing big data. It would encompass the fields of criminology, mathematics and statistics while gathering data on times, places, weather and geographical conditions as well as other aspects of crimes and accidents.

A panel appointed by the Metropolitan Police Department to study how to make use of information and communications technology has conducted hearings with experts including Kajita.

In April, the panel, led by Takushoku University professor Tadashi Moriyama, proposed the promotion of big data, saying predictions are expected to help prevent crime and improve security.

The system is expected to be initially used for police patrols and by citizens. Based on that, more comprehensive information and communications technology (ICT) for security purposes is expected to be put to use during the Olympics. ICT could also help forecast congestion at major events, the panel said.

The panel also highlighted relevant problems, such as the handling of personal data.

Among the challenges that implementing such a system would bring, the panel cited the task of training police to use ICT on the job, keeping the systems secure and protecting personal information.