Press Release - 5th August 2009

7.5 Press Release - March 2010

Chris Bates is senior lecturer in software engineering and technical lead on the Odyssey project.

Odyssey is an attempt to find a way of unifying data about gun crime from police forces across Europe. There's lots of gun crime across Europe, ranging from homicide, through incidents where a gun is waved around but not necessarily fired, through to incidents including discharge of a gun. The guns that are used in crime in the UK have typically come in from overseas, so there's a need to track guns back and see patterns in crime and movement of guns. At the moment there's no way of sharing that data, so Odyssey is an attempt to show how that data can be shared.

We've got partners from Slovenia, Italy, Spain, Belgium, Ireland and the UK. It's a big, multinational project, with a diverse range of cultures and requirements within the partnership. There are two types of partner involved - those who are involved in the development of the system, and the end-users of the system. Those include the Italian Police, the Royal Military Academy in Belgium, the Gardia in Ireland, West Midlands Police, North Yorkshire Police, the Met and the National Ballistics Centre in the UK. Those partners are on board as advisers about what end-users want from the system - what they would do with it, how they would use it.

What the end-users really want to be able to do is to take a bullet, a cartridge case or a gun, and know that the gun involved has been used in other crimes and that there's a detective at the other end that they can speak to about it. They don't necessarily want to know what other crimes it was used in, but the key thing is having someone available to speak to. For example, say a gun is used in an armed robbery in Germany, but it may also have been used in a crime in Poland. At the moment the only way these two crimes could be connected is if the two detectives actually met and talked about the crimes.

What we're trying to do is find the data about the crimes and the guns and flag up to the detectives involved in the separate investigations that the same gun may have been involved in both crimes, and then put them in touch with each other. We're not automating any of the investigative process - investigations are a very human thing, reliant on the skill, knowledge and approach of the detective. You can put technology there to support that but there's a danger that technology can actually get in the way. What Odyssey does is enable detectives to talk to each other as simply as possible.

Is there anything already out there like Odyssey?

No. The way agencies test-fire guns produces an image of the bullet with patterns of marks on it. That approach is common and there are a number of companies which make those machines. In the US, all law enforcement agencies use the same make and type of machine and so those machines can, in theory, all talk to each other. In Europe, every jurisdiction uses a different type of machine. So they can't talk, there's no standard machine and you can't take the image from one machine and compare it with one from another.

Another issue is the standards associated with crime data. Do we record things about the criminal? How is data on a body recorded? What time data do we record? Is it the time the crime happened, or the time it was reported? In some forces in the UK there are standard approaches to these things. But there are 43 forces in the UK alone and loads more across Europe, so we've been looking closely at how those forces record crime data. We're trying to create standards specifically around ballistics: bullets, cartridge cases, shells and guns. We're saying 'in an ideal world this is the data you should collect, so that it can be shared'. Not the scratches and burn marks that occur on a bullet when a gun is fired - that's already recorded by the machines. But how do you record things like the calibre of the bullet or what it's made from? Is it a genuine bullet or is it made in an illegal factory? We're looking at how agencies record that information and we're going to standardise some of those things.

The project's now in phase two. What's new?

The first year was about understanding the field of play, understanding the users and the problems. Now we're starting to build the computer architecture and looking at how we get the machines to talk to each other, and making that process secure. We're now thinking about how we take the bullet data from crimes and process it so we can see where things match. We're using two technologies to do that. One is data mining, whereby you process large amounts of data using standard algorhythms, and this reports patterns back to you. The software we're using for this is SAS, and they're a key partner in Odyssey.

We're also using techniques around semantics and the meaning of the data. Hopefully, one of our breakthroughs will be that we can take the data mining stuff and the semantic, meaningful data and apply probabilities to it. For example, if there are reports of two crimes we should be able to say how similar the patterns of the crimes are. We can then cross-reference that with witness statements, or other relevant data.

The next step is to look at the intellectual property, and the idea is to go to market with products based on Odyssey and based on the standards that come out of it.