London police trial gang violence 'predicting' software

  • Published
Gang membersImage source, Thinkstock
Image caption,
Accenture says the software identifies high risk groups rather than specific individuals

Police in London have tested software designed to identify which gang members are most likely to commit violent crimes.

The 20-week pilot study is thought to have been the first of its kind in the UK, although similar experiments have been carried out elsewhere.

It used five years worth of historic data, but the idea would be to analyse up-to-date details if it is deployed.

Civil liberty campaigners have voiced concerns.

But Accenture - the firm that developed the software - highlighted the potential benefit it offered.

"You've got limited police resources and you need to target them efficiently," said Muz Janoowalla, head of public safety analytics at the company.

"What this does is tell you who are the highest risk individuals that you should target your limited resources against."

Flagging threats

The software works by merging together data from existing systems already used by the Metropolitan Police and carrying out predictive calculations.

Types of information ranged from previous crimes to social media activity.

"It's previous offending and various different sources that are used for intelligence, in terms of who they are involved with and who they associate with," explained Sarah Samee, a spokeswoman for the Met's Trident Gang Crime Command.

Mr Janoowalla added: "For example if an individual had posted inflammatory material on the internet and it was known about to the Met - one gang might say something [negative] about another gang member's partner or something like that - it would be recorded in the Met's intelligence system.

"What we were able to do was mine both the intelligence and the known criminal history of individuals to come up with a risk assessment model."

Image source, Getty Images
Image caption,
The Metropolitan Police Service could benefit from "big data" analysis of gang members

The study used data gathered about known gang members across London's 32 boroughs across a four year period to forecast their likelihood of committing further violent acts.

This was then compared to known acts of aggression that took place in the fifth year to give an indication as to whether the software was accurate.

Mr Janoowalla said the intention was to identify groups of gang members that were at the highest risk of reoffending rather than singling out specific individuals.

He said that he was confident the experiment had been a success, but added that he was not allowed to disclose the exact criteria on which the software was being scored.

Image source, Thinkstock
Image caption,
Big data analysis is designed to make connections between different databases of information to deliver new insights

Privacy campaign group Big Brother Watch has asked for more information to be made public.

"The police need to be very careful about how they use this kind of technology," said research director Daniel Nesbitt.

"Big data solutions such as this can run the risk of unfairly targeting certain groups of people and potentially making them feel stigmatised as a result.

"The Metropolitan Police must ensure that they are fully transparent about how they intend implement this technology and what type of information will be used in the process."

In response Mr Janoowalla noted that the Ministry of Justice already operated the Offender Assessment System and Offender Group Reconviction Scale (Oasys) - a computer-based system used to predict the likelihood of different types of released criminals reoffending.

He said the key difference with Accenture's software was that it was specifically tailored to tackle gang violence.

Data-driven policing

While Accenture and the Met believe this is the first test of its kind in the UK, the company has carried out other crime-prevention analysis elsewhere.

In Spain it has tried to identify locations where crimes are most likely to happen, and in Singapore it has tested software that monitors video feeds of crowds, traffic and other events to alert the authorities to potential risks.

Other companies are pitching rival tools. IBM has explored how factors including weather patterns, past crimes, and surveillance efforts can be combined to predict threats.

And police in Kent, Greater Manchester, the West Midlands and Yorkshire have all trialled software from PredPol, a US start-up, to help tackle street crime.

Image source, Thinkstock
Image caption,
Campaigners worry that big data crime prevention efforts could be a slippery slope

However, campaign groups have warned against the danger of police gathering too much personal data.

"It is clear that harnessing and analysing vast data sets may simplify the work of the police," said European human rights group Statewatch earlier this year

"However, this in itself is not a justification for their use. There are all sorts of powers that could be given to law enforcement agencies, but which are not, due to the need to protect individual rights and the rule of law - effectiveness should never be the only yardstick by which law enforcement powers are assessed.

"The ends of crime detection, prevention and reduction cannot in themselves justify the means of indiscriminate data-gathering and processing."

Related Internet Links

The BBC is not responsible for the content of external sites.