Cyber-detective assists police with criminal investigations
EU-funded researchers have been developing next-generation analytics technology that will help law-enforcement agencies make sense of the masses of data collected during complex investigations. This will help to increase crime-detection rates and improve security for citizens.
© Rick McCullagh, 2017
While investigating serious crimes such as terrorist attacks and child abuse, police forces often struggle to cope with the volume, fragmentation and complexity of evidence they collect and process.
The EU-funded VALCRI project has been working with police forces to develop a series of data-processing, analytic and sense-making tools that will help authorities determine which information may be relevant or useful and help investigators make sense of the mass of data.
As one smartphone could hold over 30 000 pages of text or images, it is not humanly practical to sift through such enormous volumes of data, says William Wong of Middlesex University in the UK, coordinator of the VALCRI project. Some police forces in Europe are dealing with 250 different databases. When a crime occurs, all you have are bits and pieces of out-of-sequence information. Investigators have to combine information from crime reports and databases to construct an idea of what might have happened and how.
How analysts think
Police forces already compile statistical analyses of crimes to look for patterns, but this may not be helpful for active investigations into offences such as human trafficking or child sexual abuse. You are trying to find hidden threads, says Wong. Who are the pimps, who are the couriers, where are the safe houses, how do they transport people around? It is difficult to solve such problems if we mainly summarise data.
VALCRI was not designed to automate the assessment and decision-making currently done by police analysts. The aim was to support how analysts think, rather than what analysts do.
A key function of the project was hypothesis testing. Analysts use expertise and intuition to form a hunch, a plausible explanation that accounts for the known facts and may suggest a suspect. VALCRI has helped analysts to formulate and test alternative hunches by rigorous application of scientific method, minimising the impact of cognitive biases that affect human reasoning.
It helps investigators find, collect, sort, assemble and reassemble data into explanatory sequences, says Wong. If one can construct a hypothesis in five minutes rather than five days, one is more likely to discard unsuitable hunches and start again.
While ethical issues could arise from trawling through large amounts of sensitive data, plus the risk of innocent people being arrested and questioned, Wong thinks the answer may lie in computational transparency. Here, VALCRIs chain of reasoning is open to inspection, so the decision to act is made by someone who is accountable.
Over four years, 17 partner organisations, including three police forces in Belgium and the UK, have been working on more than 75 tools comprising the VALCRI system. The prototype has been tested with anonymised data supplied by West Midlands Police in the UK, and police trials involving real data have recently started in the USA and Europe.
Two of the SME partners have produced training courses on intelligence analysis, including a Masters-level course at the UKs Aston University.
Another has obtained a patent on part of the VALCRI research, while another is developing an information-retrieval system for police in Belgium.
Since the project ended in the summer of 2018, a commercial partner has taken over the development with the intention of bringing VALCRI to market.
Not only will this create new jobs in Europe, Wong says, but commercialisation will make it possible for police to acquire the next generation of sense-making analytics technology for crime investigation and intelligence analysis.