West Midlands Police, the second largest force in the UK, is seeking to reduce risk to its officers and the public by using software that gives a master record for criminals and suspects.
Corinne Brazier, force records manager at West Midlands Police, will tell delegates at IRM UK’s Data Management and Information Quality Conference in London next week about the data quality efforts she and her team have been making since 2009, including the implementation of a “golden nominal” master record for persons of interest.
“Our ‘customers’ don’t always want us to have correct data about them,” she says, and criminals will often exploit the information management challenges the police must face.
Data capture in police work is not easy. Front-line officers are often in high-pressure situations. Staff taking 999 calls can make typing mistakes. And even data entered at relative leisure can be rendered unsearchable if it is put in the wrong field or misspelt.
Careful data management builds criminal profile
West Midlands Police serves a population of more than 2.6 million people, covering Birmingham, Coventry and Wolverhampton. It employs almost 8,000 police officers and 3,500 support staff.
West Midlands Police has achieved a “golden nominal” master record, aggregating the variety of records the police have on each person of interest
It has a home-grown information viewing system – Flints (Force linked intelligence system) – which was developed in 2000 and is fed by 15 databases.
“But because of data quality problems we were not able to build up one picture of each person,” says Brazier. “Duplicate [records] are the bane of a police officer’s life.”
Previously, she says, the data was usually there to be found – if you had the time to search in various different ways.
And then there were the information linking requirements of the Management of Police Information (MoPI) initiative that came out of the Bichard Inquiry into the murders of Holly Wells and Jessica Chapman, in Soham, a decade ago.
Brazier’s team has achieved a “golden nominal” master record, aggregating the variety of records the police have on each person of interest. The team aggregated 14 million records into four million consolidated views, using DataFlux Power Studio, bought in 2009 and fully implemented by the MoPI deadline of 31 December 2010.
Since then the force has been working with the golden nominal system, alongside BlueStar's Corvus operational briefing system for police officers and support staff.
Brazier would like to do more with the DataFlux technology. One next step would be a “golden address”. Another would be to enhance the golden nominal with last photograph, last address and similar information. And, above all, get real-time correction of data input errors.
Read more on data management in law enforcement
- Law enforcement data mining, predictive analysis: get this book excerpt
- Case study: Analytics take a bite out of crime
- Text analysis software, content analytics help fight crime
Accurate criminal records essential for effective policing
It is a common misconception, she says, that back-office staff are less relevant to real police work. “Back-office staff, such as ourselves, provide the data quality audits that ensure the searchability of information. That reduces risk for front-line officers.
“Information like the golden nominal gives police officers a clearer understanding of what risk a suspect might pose. Do they have a history of knife crime? Who are their associates? What do we know about them?”
She says, however, that potentially less than half of the 43 police forces in England and Wales are using a golden nominal system.
Brazier has presented the team’s data quality work to the top-tier leadership groups in the West Midlands force, which are very supportive.
“That is always the tough thing [in data management] – how to get senior management buy-in.” And, lower down the organisation, it is a matter of relating data quality to officers’ and support staff’s roles.
“Talking to front-line officers, I try to relate it back to Soham. So I stress to be careful to get the data correct, and be aware of what other database sources could be helpful,” she concludes.