Computer program reveals where criminals are most likely to hit
By Lucian Constantin
[softpedia.com] British police forces are trialing computer software which is able to pinpoint potential criminal hot spots from data fed to it. The system leverages statistical history and evaluates patterns to predict the places where crime is most likely to occur. Known as CRUSH (Criminal Reduction Utilising Statistical History), the program was developed by IBM with help from the University of Memphis Criminology and Research department.
The system has real time access to a wide variety of data including, but not limited to crime reports, police intelligence, offender profiles or weather forecasts, which it combines with past crime patterns to accurately pinpoint crime hotspots. The system was used successfully by the Memphis Police Department for a long time, where it allowed resources to be allocated efficiently to decrease law enforcement response times.
Such proactive actions are said to be responsible for a crime reduction rate of 31% in the area. “This is more of a proactive tool than reacting after crimes have occurred. This pretty much puts officers in the area at the time that the crimes are being committed,” John Williams of the Memphis Crime Analysis Unit, commented for The Guardian, which also reports that two British police forces have secretly begun trialing the software. The methodology behind such programs is known as “predictive analytics” and encompasses various data mining and statistical analysis techniques. The technology has long been used in certain industries, particularly financial services, and in various fields of research. However, applications in the justice system, like for determining which prisoners are most likely to become repeat offenders has been strongly criticized by human rights activists. IBM explains that the CRUSH system does what police officers were already doing, but on a whole new level and with much more speed. Current development efforts aim to increase the number of sources the program can process, with CCTV camera feeds and Facebook posts being strong candidates.