Artificial Intelligence System Designed to Help Detectives Close More Cases in Less Time

Police detectives devote their professional lives to solving cases and getting criminals off the streets, but even they are unable to know about every report filed. Detectives also develop a keen eye for determining patterns of criminal activity, but without every bit of data in front of them, it is difficult and time-consuming for even the most seasoned police veteran to spot potentially related crimes.

However, researchers at DePaul University's School of Computer Science, Telecommunications and Information Systems (CTI) have developed an artificial intelligence system that is designed to spotlight potential crime patterns so detectives can close cases more rapidly and put more criminals behind bars.

The system, called the Classification System for Serial Criminal Patterns (CSSCP), was the result of several years' worth of research initially funded through a National Institute of Justice grant. Using an artificial intelligence program called a Kohonen neural network that trains itself to detect patterns, the CSSCP scans records of criminal cases, sorting them by characteristics (such as type of offense, weapons, vehicles involved, etc.) and descriptions of the perpetrator and victims. Through these profile statistics, it looks to match other crimes with similar profiles. A list of possible patterns is then provided to detectives to investigate further or discount.

While there have been other computer profiling programs put to use by law enforcement agencies in recent years, what makes the CSSCP unique is that it will search for patterns without the prompting of a human operator or programmer. This will enable police to pinpoint patterns of serial criminal activity more quickly as they emerge, rather than long after the pattern (and crime) has been completed.

In a trial using armed robbery statistics from three years of incident reports, the CSSCP was able to correctly classify patterns at a rate many times higher than the one percent of serial case closings that a Rand Corporation study notes is more commonly attained. While most detectives would eventually identify all of the patterns spotlighted by the system -- if given the time and resources to track all cases -- it would be time and cost prohibitive for them to do so.

"This shows that detectives using the program could possibly close many times the number of cases that they currently are closing," said Tom Muscarello, associate professor of DePaul CTI and one of the study's lead investigators, along with CTI colleague and assistant professor Kamal Dahbur. "When you arrest someone for armed robbery, you may get him to admit to one or two other cases, but you're not going to get him for the 30 or 40 instances he may have committed but isn't telling you about. But with that pattern more accessible to police, they'll be able to close more of those cases. It's not that the detective wouldn't identify the pattern on their own, it's just that they don't have the time to sift through the mountains of information quickly enough."

What the system provides more than anything, Muscarello added, is a starting point for law enforcement investigations that saves time and resources for cash and manpower-strapped agencies.

More importantly, the CSSCP system is not an expensive investment for most police departments, requiring only a computer terminal with a powerful processor, such as a UNIX server or a small PC running Linux. The program can run on its own during non-peak hours, to avoid clogging the department's computer network or work stations. Muscarello is currently seeking to set up live trials of the system with the Chicago Police Department, as the system was developed with the input of several police detectives in the city. He is also looking at potential applications of the system in preventing terrorist incidents.

"Data is data," Muscarello said. “If the program can spot patterns of crimes, it can spot patterns of potential terrorist activity."

DePaul CTI is one of the most innovative and wide-ranging computer science programs in the country. The undergraduate program enrolls 1,200 students and offers 11 different degrees. More than 2,130 students are enrolled in its 17 graduate programs. CTI also features a doctoral degree program in computer science.

With a total enrollment of 23,570 students on two city and five suburban campuses, DePaul is the seventh-largest private university in the U.S. and the largest Catholic university in the nation.