Data correlation helps recognize pickpockets

In the fight against mobile banditry, the police, together with Eindhoven University of Technology, are going to do a trial with data correlation. By cleverly combining data from, for example, number plates, camera images and messages on social media, the deviating behavior of, for example, pickpockets can be detected more quickly. The cooperation between the two parties will start officially on Wednesday 11 July.

The term mobile banditry is used for international itinerant criminal groups, which are guilty of various crimes, including pickpocketing. They often do this in busy places such as shopping centers, exhibiting the same patterns and moving through the shopping centre differently from the normal shopping public. The combination of the available data and the application of innovative data analysis techniques should lead to better and faster recognition of this deviating behavior, so that the gangs can be stopped even before they have been able to strike in the shopping center.

Trial
One of the places where data correlation is used is in the center of Roermond. With tens of thousands of (international) visitors every day, the city is struggling with a serious pickpocketing issue. A university team led by TU/e data-mining professor Mykola Pechenizkiy will assist the police with the analysis of various databases containing information about visitors to this shopping center, including, for example, automatically recognized number plates, data collected for marketing purposes, messages on social media and camera images.

Sensing program
The trial in Roermond is one of the local living labs within the so-called Sensing programme of the National Police. The aim is to test the new possibilities of using data from, for example, cameras and other sensors in a responsible and controlled way before they are used in practice. This collaboration with TU/e is about developing new data analysis techniques to speed up the recognition and prevention of criminal behavior.

Privacy
Naturally, privacy issues form an important part of this project. By trying out new techniques in living labs, their impact in practice becomes clear. Not only for the criminals the project is aimed at, but also for the other visitors to the shopping center. Marius Monen of the TU/e Data Science Center Eindhoven explains. “For this reason, we also work closely with the mayor and public prosecutor, for example. Based on the experiences in the living labs, they can assess the impact on the privacy of citizens and decide whether this is acceptable in order to achieve the desired goals.”

Other living labs
This project was partly the result of a cooperative effort between the police and the DITSS (Dutch Institute for Safety & Technology).  It is possible that the cooperation between DITSS, the Police and TU/e (working with Tilburg University in Jheronimus Academy of Data Science, JADS) will be expanded in the future. Elle de Jonge, a chief inspector of the police force and national coordinator of the Sensing living labs: “We expect to be able to use the analytical techniques that have already been developed in other living labs that are still under development. For example, to prevent the blowing up of ATMs (‘plofkraken’) and to recognize deviant behavior in shipping to intercept drug runners.”

Source: Data Science Center Eindhoven, website, 11 July 2018.

Pickpockets