• 08.04.11

The Warehouse Where No Crime Is Forgotten

IBM’s predictive Crime Information Warehouse (CIW) technology collects data and spits out real-time, vital information for investigators. Here’s how cops are using it to stop crimes before they’re committed.

The Warehouse Where No Crime Is Forgotten

A new crime-busting warehouse is helping officers spot crime trends as they develop in real time. IBM‘s Crime Information Warehouse (CIW) is a lean, mean data-mining machine that puts all the little pieces of the puzzle together to help police anywhere connect crimes, spot trends, and crack cases faster, sometimes even before new crimes happen.


Police departments excel at capturing information, but there’s no efficient structure for dealing with where all that crime data goes afterwards. Critical case information ends up sitting in pockets across many departments and it’s left to officers to tirelessly track it down. That’s less time spent doing what they were trained to do–solve the crimes.

That’s where the CIW steps in. The software solution currently being applied in Richmond, Virginia, New York City, and Edmonton, Alberta, Canada, acts as a repository for crime related data, pulling information from multiple standalone systems to give officers a one-stop access point for all their data needs. Integrated crime statistics made available in real time can revolutionize policing tactics, making it “predictive”–if officers can see the trends forming through patterns in the data, they can more likely prevent any further similar crimes.

“Analytics recently identified a neighborhood as having a spike in thefts from vehicles,” says John Warden, Manager, Business Performance Section, Edmonton Police Service. “Officers were assigned to go door to door in the area canvassing the occupants for further information. They spent about four high-visibility hours talking to the community residents. No one was arrested or identified as a suspect, but there were no further thefts from vehicles in the area for three weeks.”

Data-driven crime prevention began with the introduction of CompStat (computer statistics) by the NYPD in 1993. A multilayered initiative, it employs geographic information systems to generate crime maps that officers can use in weekly meetups to employ new strategies to cut crime. Credited with decreasing felony rates in New York city, the program is successfully being used across many agencies in the U.S. to address spikes in crime.

The CIW [PDF] adds the critical element of speed, mining data in real time to provide a holistic view. IBM developed the solution to synthesize bits of data into actionable intelligence; using GIS mapping, advanced software, and visualization tools, the warehouse can clean, gather, link and sift through large amounts data to discover correlations and patterns, delivering it directly to officers in the field.

Having a tool to do all the grunt work speeds up things. For instance, a detective looking for a person at a location could be sent data on any other people living there, their connection to the suspect, criminal history if any and previous callouts to that address. An officer could get an email alert when a person concerned with an investigation is arrested elsewhere. Forensically, the solution could help reconstruct events to figure out what went on in the area in the lead-up to the crime.

The key here, says Mark Cleverly, Director Of Strategy, IBM’s Global Government Industry, is the warehouse’s ability to allow officers to drill into a zone of interest and analyze historical arrest information, giving authorities the power to understand where to deploy forces and why.


“Key attributes such as offense, time of day, day of week, dispatch zone, weather conditions, average temperature, event types, as well as occurrence of a violent crime incident in the past 24 hours and past week are captured to identify circumstances under which a violent crime has occurred,” Cleverly says. When predictive techniques are applied on this historical data, it renders forecasts on a color-coded map for each dispatch zone, depicting the likelihood and intensity of future incidents.

Seeing data in a meaningful way is a critical step forward toward shifting from reactive to proactive policing–matching tactics against problems. If the data shows a 30% increase in robberies, for instance, the officers can dig into it to see that it represents 14 more robberies, of which 80% are street robberies. The CIW can track the results of any immediate action taken to assign resources in those troubled neighborhoods to see if the tactics employed really did reduce crime rates.

“The goal is to begin identifying trends earlier based on information that may not have an obvious impact on crime, but may provide a piece of the puzzle to identifying the crime before it actually takes place,” says Dale Peet, a crime trends expert from the Memex solution team at SAS. “Criminal acts such as terrorism often give strong early indicators that could lead to prevention.”

That’s providing the warehouse can be deployed as a solution relatively easily.

“This is not by any stretch of the imagination a plug-and-play item,” says Nick Selby, managing director, CSG Analysis and a Texas police officer. “It requires a degree of organizational and technical maturity, and a cogent IT strategy, which by definition limits its applicability to only a handful of agencies in the world. While aspects of it would be applicable to local law enforcement agencies, it’s probably overkill–it’s swatting a mosquito with a Howitzer.”

IBM officials say the CIW is in any case customized to suit a specific client. “Cities and geographies differ enough that models should be created and updated relative to their specific circumstances,” says Cleverley.

How far can it go eventually? As the system develops more links with video systems that monitor urban scenes, the CIW could generate alerts in close to real time for little acts like objects being left unattended, vehicles performing odd maneuvers, etc. A crime-tracking big brother in the making, perhaps? That’s one prediction.


[Image: Flickr user diveclimbsurf]