Police want to predict crime with data platformNational police are investing 5.24 billion won ($4.79 million) to develop a big data platform to investigate crimes. Investigators will be able to share clues from different cases - and even predict crimes in advance.
The platform, called Crime Layout Understanding Engine (CLUE), was planned and developed from February 2016 and is expected to be launched in early 2019.
The Korean National Police Agency said it is coordinating with IBM, Seoul National University’s Statistical Research Institute, a big data company and a language-processing software company to create the system.
CLUE will be a digital platform on which all investigation records are pooled, and police can search cases by keywords and specific terminology used by police. This is where the language-processing software comes in, police said.
By comparing similar cases through CLUE, police expect they could gain a head start in an investigation or analyze the motives of a criminal faster by comparing similar characteristics in past crimes. The platform will also include information on the demographics, real estate price and even weather conditions per region in the country to try to predict where certain crimes are more prone to take place.
“The platform will use an algorithm that can comb through all the information out there on the internet in addition to the information inputted by police,” said a police spokesman. “Using all this information, it will try to predict where crimes could take place.”
Reminded that the movie “Minority Report” dealt with the dangers of a crime prevention system, the spokesman said, “CLUE does not use personal identities and records. It will not be designed to keep a file on individual identities. And CLUE’s analysis carries no legal weight. The police simply will use its analysis as a reference point in an investigation.”
“The system will also be secured through an eight-stage security system so it will be safe from hacking,” he added. “And the system will only be used by police.”
Other programs to predict crimes have existed before. The New York Police Department’s CompStat, created in 1994, is a program that uses crime records to predict areas where crimes are most likely to take place every morning.
In 2011, police in Santa Cruz, California, started using a program that analyzes patterns of crimes per area in the city to try to predict what types of crimes may take place in an area the next morning. With the adoption of the system, the area saw a 19 percent drop in thefts, according to PredPol, an organization that focuses on so-called predictive policing.
In 2015, the Chinese government implemented a system called Tianwang that connects 20 million CCTV cameras throughout the country to a database of criminal suspects. The idea is that if a camera catches people running or cars speeding suddenly, it will analyze their faces and run them through the criminal database. If any of these faces match, the police are alerted. The Chinese government is also said to be developing a program named Isvision, through which anyone could be identified within a few seconds by his or her face caught on an artificial intelligence-equipped camera. The U.S. government was also reported to be looking into developing an AI program to vet immigrants.
BY SONG WOO-YEONG, KIM JUN-YOUNG AND ESTHER CHUNG [email@example.com]