Toronto police testing IBM analytics software

LAS VEGAS – The Toronto Police Services is nearing the completion of a proof-of-concept project with IBM Corp.’s analytics and name recognition tools.

The TPS was in attendance to share their initial experiences with IBM’s suite of tools at this year’s Information On Demand 2009 conference in Las Vegas, the company’s annual business intelligence and data management show.

The police service outlined its use of IBM’s Entity Analytic Solutions toolset, which is intended to help firms recognize who they are interacting with using real-time identity and relationship recognition capabilities.

In addition to EAS, the TPS has begun implementing IBM’s InfoSphere Global Name Recognition platform, which helps with name-matching across their analytics applications. The GNR tool is especially useful for multi-cultural name-matching, according to Dave Angus, an officer responsible for managing several crime databases at the TPS.

While Toronto’s police department is still in the proof-of-concept stages with both products, the organization has seen huge benefits, Angus said.

The TPS has been able to do things such as monitor all the people out on bail, track whether or not the same person has been bailing out multiple suspects, and pull together unstructured and buried data in police narrative reports, he said.

Using the EAS tool, the police department also expects to ramp up its work in predictive analytics. Angus said the TPS is already tracking information such as criminal behaviour patterns, average sentences and other data to bolster their court cases.

“We can now go to the courts and have some weight behind us,” he said. “We can say that this type of person is unlikely to show up in court and have the data to support it.”

Similar work is being done at the New York Police Department under the watch of Ruben Beltran, a deputy chief and the head of the NYPD’s Office of Information Technology. Whether it be monitoring indirect relationships, keeping track of aliases, or even bringing in parking ticket data, Beltran said that EAS is helping its officers get a clearer picture of suspects and a strong tool in tracking down witnesses.

Earlier this month, the NYPD’s EAS system helped solve the murder of a 92-year-old women who was hit by a stray bullet while watching TV in her apartment.

“Investigators doing the canvass discovered that the suspect went by the nickname ‘Fish,’” Beltran said. “Using EAS, we were able to find ‘Fish’ as an alias connected to the suspect, we arrested him and got a confession.”

In addition to nickname databases, the NYPD is even heading toward a tattoo database, as the popularity in that art form continues to grow.

For the TPS, the ability to receive alerts within the EAS reader, Angus said, will be crucial to keeping tabs on potentially dangerous individuals and their activities.

“We monitor people likely to be involved in gun activities and usually have to tick them off the list when they either get shot and murdered,” he said, adding that in the past officers would simply scan the newspaper to see if a potentially dangerous individual was killed.

But without real-time alerts connected to other disparate systems in the TPS, Angus said, the police department would not be aware if these potential offenders died in a house fire or car accident. The EAS tool will allow the TPS to automatically manage these lists and take advantage of all their analytics systems.

“I can also look up the top 10 most violent people that aren’t in custody,” he added.

As for the biggest drawback of using both the EAS and GNR platforms, Angus pointed to IBM’s steep asking price, which he said could price the tools out of range for smaller law enforcement agencies.

An area that both the New York and Toronto police representations want to see more work done on is multi-cultural name recognition. Both cities are very culturally diverse and not every suspect has a clear first, middle and last name.

While this functionality was a selling point for both police departments, with hundreds of languages and dialects spoken in each city, any improvements to this area will be welcomed.

For Frankie Patman, manager of the linguistics department at IBM’s global name recognition division, the fact that public sector organizations such as hospitals or police stations often have to force names into those database fields is troubling.

“We’re never sitting still on this and obviously we want to cover names for more parts of the world, especially countries that have writing systems that aren’t Latin-based,” she said.

The speed of name recognition searches and the user interface of the software are another two issues that will be continually addressed by the GNR team, she added.

While EAS and GNR tools are helping both police departments handle the real-world information they are collecting and bringing into their data systems, keeping tabs on the Internet is a whole different story.

At the TPS, Angus said a proof-of-concept has been issued in regards to a Web-crawling tool, while Beltran indicated that the New York police have yet to find a tool to solve this issue.

Currently, both organizations are scouring the Web for information manually.

This is a practice that actually helped Toronto police control summer protests over the war in Sri Lanka, Angus said.

“We had people actively monitoring Twitter and Facebook, with protests communicating when they were going to meet and where,” he said, adding that sometimes the messages were being posted in code.



Related Download
IDC Marketscape Worldwide Virtual Client Computing Software Sponsor: VMware
IDC Marketscape Worldwide Virtual Client Computing Software
For a qualitative assessment of the characteristics that explain a vendor's success in the marketplace, download this IDC Study.
Register Now