Cops and super spies on television and in the movies use technology to capture the bad guys all the time. They prove themselves to be whizzes at search and make connections timed perfectly to the storyline. But more and more, real crime fighters and security experts are using text and content analytics to collar criminals.
For example, the Richmond (Va.) Police Department recently investigated a series of robberies during which victims were directed to lie on the ground. A pattern emerged in the phrase the robber used when addressing victims. Investigators searched on the phrase using a pilot version of text analysis software and connected a series of incidents that helped police establish the suspect’s modus operandi, leading to his arrest.
In a major East Coast city, whose police department asked not to be identified, an investigator searched on a phrase tattooed on a crime suspect’s arm. The search, using natural language analysis, turned up another investigator’s description of the same tattoo taken from a witness to an unsolved murder years before. Police connected the suspect to the old crime and closed what otherwise might have been a cold case.
And in North Carolina, police are using content analytics to better distinguish between the good guys and the bad guys to protect themselves and improve investigations.
Cost and expertise remains an issue
“Natural language search has huge promise for police agencies,” said Colleen McLaughlin McCue, a consultant currently working with the departments of Defense and Homeland Security on using text and predictive analytics to stop domestic and international threats. She said local police department adoption of the technology has “been very close to the tipping point for a long time,” but while most departments have the ability to do key word search, cost and existing skills “are holding departments back.”
Effective text analysis requires some linguistics expertise. While many private companies can pay to have linguists on call or train an employee to be a language expert, municipal governments can rarely justify the expense. In an effort to help out, McLaughlin McCue, who has a doctorate in psychology, said the federal government was creating fusion centers to make such technology available to other agencies. But there’s no word yet on when text analytics might be included.
A game-changing technology for national security
The technology is, however, playing a key role in national and international security realms, McLaughlin McCue said. “The narrative section is where the good information is. The ability to use natural language search can be game-changing.”
She said the Department of Defense was using text analytics to search communications overseas to gain a better understanding of the environment, culture and attitudes among the populace in a particular hotspot. It can “be very effective in helping us work with the local population,” McLaughlin McCue said. “I think it’s getting much more intuitive to use. … It really supports exploration and discovery.” She added, though, that the databases of vocabulary need to be more “law-enforcement specific.”
Written descriptions of tattoos and graffiti, as well as image data, are being stored by police agencies, even if they do not have text analytics capabilities, McLaughlin McCue said. And those that do are “reviewing the content from a semantic perspective and modeling is being done to identify trends and patterns in series of crimes.”
Content analytics keeps police safe and focused
Content analytics, on the other hand, is being used by police in North Carolina, according to Carol Burroughs, the project manager for the state comptroller’s Criminal Justice Law Enforcement Automated Data Services (CJLEADS) program, which was deployed to 8,000 users in January.
The system, which is currently being used by police, court officials, juvenile court counselors and probation officers, requires a three-hour training and automatically searches court and parole records, the state’s sex offender registry, arrest records and more. Hunting license records and recreational vehicle registration information will be added to the system as well to help build a citizen information database.
The idea, Burroughs said, is “to paint a picture of a person to find out whether he’s a good guy or a bad guy.”
Before CJLEADS was deployed, when an officer entered a license plate number into a patrol car’s laptop, the only information returned was a name and any traffic violations or suspensions tied to that number. Now, a picture of the registrant is returned along with a set of icons indicating whether the person has been noted to be violent, a court escapee, a sex offender, on probation or whether he has been associated with weapons, drugs or gangs. The photo also saves time and effort when an officer pulls over someone who’s forgotten a driver’s license.
Program as data management warehouse
Burroughs said she views CJLEADS as an integrated warehouse of data managed by advanced analysis and search capabilities. It will be expanded to include geospatial data later this summer with the aim of combating child abductions, and the hope is to someday add facial recognition capabilities.
The system’s been very popular, and potential users have been calling to find out when they can get started with it. Burroughs said it would be rolled out to another 22,000 users by next summer. And aside from warning police to approach someone with care or leading court officials to make better decisions about keeping someone in jail, it has led police to help Alzheimer’s patients and others find their way home.
“It isn’t just about criminals; it’s also about citizen safety,” she said.