Plotting a path through crime data Vast amounts of data are available to the police in the fight against crime, yet it is not always easy to sift through it and identify key points. We spoke to Professor William Wong, Dr Chris Rooney and Dr Neesha Kodagoda about the VALCRI project’s work in developing an intelligent system to support police analysts and help them work more effectively The
police and law enforcement agencies today have access to large volumes of data in the fight against crime, yet sifting through it and identifying the key points relevant to a specific incident or tactical policing operation can be a challenge. The effective use of intelligence and historical data can help police analysts gain important insights when investigating crime, a topic that lies at the core of the VALCRI project. “We aim to develop a simple and intuitive interface to help police analysts drill down into what sorts of crimes are interesting with respect to a certain investigation. We can then provide additional support on the similarities between these crimes, and which of them show certain attributes,” explains Dr Chris Rooney, the project’s Technology Lead. This builds on the knowledge that criminals can be predictable in the traits they reveal and the way they operate,
72
their modus operandi (MO). “They may like to enter a property in a certain way for example, because they know that it works for them. So it’s about detecting those specific traits and identifying other crimes where those traits have also occurred,” says Dr Rooney.
Reasoning workspace A lot of this kind of data has been amassed over the years, so a key challenge now is ensuring it can be navigated effectively and presented to police analysts in an accesssible, easy-to-digest way. This is not always the case at the moment, as it can be difficult for analysts to locate and identify the information relevant to an investigation. “Data is often spread across different systems and databases, so it can be quite hard for analysts to piece the data together,” outlines Dr Rooney. The project aims to help address this issue by
integrating a number of technologies into a coherent working environment for police analysts, called the Reasoning Workspace, which enables an analyst to see the key information on a single interface. “Currently analysts use lots of different Screen shot of a crime card that is used to collate possibly relevant background data from many different sources into a single source. This includes (i) details about the crime and its victims and known offenders, (ii) other crimes and incident logs that occurred in a similar time and space, (iii) the modus operandi with the analyst’s search terms highlighted, and (iv) relationships to similar crimes based on extracted traits.
EU Research
systems to do different tasks. The idea of integrating this together is that you can see the data in many different forms, all within the same interface,” says Dr Rooney. “Similarly, we plan to run a repository over multiple data sets, so again you don’t need to have these different systems running for each
remove and draw in other sets of information as required, the system enables analysts to explore it quickly and efficiently, while also drawing on their own associative capacities. “We have these interactive visualisations, which we call crime cards, which can be manipulated and moved around the canvas. It gives an
We aim to develop a simple and intuitive interface to help analysts drill down into what sorts of crimes are interesting with respect to a certain investigation. We can then provide additional support on the similarities between these crimes, and which of them show certain attributes different dataset, you can bring them together into one index.” This will give analysts access to large amounts of detailed information on individual crimes, which is also regularly updated with new data that is analysed in real time within the system. If an analyst decides that particular excerpts of data are relevant to their task, they can then use the reasoning workspace to lay them out and manipulate them in a way that suits their own working processes. “The analyst is not necessarily forced into a particular workflow, they can use the space as they choose,” stresses Dr Rooney. By presenting data within an individual’s field of vision, and allowing them to
www.euresearcher.com
analyst this free-form interaction, which allows them to put data where they think it’s important,” explains Dr Rooney. “When an analyst thinks that data excerpts are related, they can then use the space to lay them out together.” The wider goal in this research is to help analysts reconstruct situations and build a deeper picture of crime in a particular area, combining their own local knowledge and experience with the available data to generate new insights and uncover possible leads. This encompasses not only operational policing and investigations of specific crimes, but also strategic and tactical analysis. “If an analyst is primarily involved in tactical analysis,
and they need to see an overview of the crime data over the past week or so for example, then they can build up a dashboard to do that,” says Dr Rooney. This type of data can be highly important in terms of strategic policing and understanding emerging patterns of crime. “Sometimes a tactical or operational analyst might be assigned a specific task, asking for example; ‘There’s a lot of knife crime going on, what’s happening? Is there a gang or is it individuals? Can we find anything similar that’s going on?’” outlines Dr Neesha Kodagoda, the project’s Operations Manager. “With the VALCRI system, an analyst can do a keyword search and generate a map, which shows them where knife crimes have happened.” The crimes recorded within the system can also be clustered together according to different attributes, as determined by the analyst themselves, helping them draw connections that might otherwise have been missed and providing an evidence base to support their reasoning. Once the key points relevant to a specific crime have been identified, analysts can then look to build a deeper picture, to effectively tell a story about the analysis that’s been conducted; Dr Kodagoda says this is an important aspect of the project’s work. “One thing we are encouraging is early hypothesis generation, so that an analyst could test a hypothesis more rigorously,” she says. The police need to put in a lot of
73
work to test a hypothesis using current systems; Dr Kodagoda says the VALCRI system is designed to help analysts develop and test a hypothesis around a crime more rapidly. “If we can make it easier to test a hypothesis, that means there is a way to check on the possible explanations behind a crime and test alternative ideas before the police commit to a specific explanation, when they need a high degree of certainty,” she explains.
Encouraging imagination This encourages analysts to consider a wider range of perspectives during an investigation, which is a major driver behind the work of the project. Many historical failures in crime investigation and detection have later been attributed to a lack of imagination in analysis, an inability to use the available intelligence effectively, a shortcoming that the VALCRI system could help to address. “The problem in intelligence-led policing is not so much the availability of data – the problem is more about trying to figure out which of the dots are useful and relevant,” says Professor William Wong, the coordinator of the project. By helping analysts draw links with historical incidents, and enabling them to drill down to gain more detailed information on specific crimes,
74
the VALCRI system encourages analysts to be imaginative and think of possible outcomes that might otherwise have been overlooked. “We aim to give the police a tool that could potentially make their lives easier and help them do their job more efficiently,” says Dr Rooney. This must not come at the cost of compromising data security however, while there are also ethical concerns around the use of this kind of information.
system is designed to support analysts in their decision-making, not to make decisions on their behalf,” stresses Dr Rooney. The longer-term goal for the project is to apply the system in policing. The project consortium includes three end-user partners – West Midlands Police Force in the UK, and both the federal and local police forces in the Belgian city of Antwerp – each of which face different
Currently analysts use lots of different systems to
do different tasks. The idea of integrating this together is that
you can see the data in many different forms, all within the same interface
Dr Rooney and his colleagues in the project are well aware of these kinds of issues. “We have a body of researchers within the project addressing SEPL (Security, Ethical, Privacy and Legal) issues, and their focus is on maintaining transparency in the system and mitigating the risks around misuse of data,” he explains. There is also a team within the project working on biases, aiming to mitigate the risk of any biases emerging and supporting objective policing. “The
challenges, and therefore require different types of data. “The strategic team in the West Midlands are primarily interested in volume data – so things like general increases or reductions in crime and monitoring any emerging trends. For example, is there a significant increase in crime in September because students are arriving in a city? How can we mitigate that? Whereas the federal police in Belgium are a bit more focused on the operational side of things, so data like
EU Research
witness statements,” outlines Dr Rooney. At the moment the VALCRI system is being tested by these end-users with anonymised data; Dr Kodagoda says the feedback so far has been positive. “Rather than looking at data in silos, they can now see this overall picture. So they have this ability to ask questions and gain answers, without having to move between different systems, leaving more time for analysis,” she explains. The system has not yet been integrated with each force’s data however, and discussions are ongoing over whether this can be achieved within the timeframe of the project. Looking beyond the funding term of the project, Professor Wong hopes to exploit the system’s commercial potential in future. “We have demonstrated the system to over 40 law enforcement agencies across Europe already, and there’s been a lot of interest in using it. We can’t sell it quite yet though,” he says. Additional funding will be required to further develop the system to a point where it can be applied in policing. “We will be at Technology Readiness Level 5 (TRL5) at the end of the project – we need to reach TRL9 for it to be considered a deployable system,” outlines Professor Wong.
Over the remaining period of the project the focus will shift towards design and implementation, as researchers move into the later stages of the development process. There is still scope for further changes and modifications however. “There are features that we’ve planned that we’ve not yet put into the system,” stresses Dr Rooney. Feedback from the end-user partners will play a large part in informing the ongoing development of the system and tailoring it to police needs. The challenges facing police in future are unlikely to be the same as those of today, so Professor Wong says that as the nature of crime evolves, so the systems used to analyse crime also need to evolve. “In future there will be a lot more information on crime available, there will be a lot more problems interfacing between crimes that occur in the physical world and crimes that occur in the cyber world. This convergence of the cyber and physical worlds will change the fundamental nature of crimes,” he predicts. “Police officers will then have to think very differently about how they bring data together in order to construct a case against a particular individual.”
At a glance Full Project Title Visual Analytics for sense-making in Criminal Intelligence Analysis (VALCRI) Project Objectives VALCRI is an Integrating Project whose goal is to develop an integrated, multi-function system prototype at TRL-5, validated in a user environment. VALCRI is developing a suite of integrated functions that are intended to facilitate human reasoning and analytic discourse. By being tightly coupled with semi-automated human-mediated semantic knowledge extraction, this will enable VALCRI to respond to human analysts in both a proactive and reactive manner, and work with analysts as a human-technology team, responding and anticipating needs as a Joint Cognitive System. Project Funding The research leading to the results reported in this work has received funding from the European Union Seventh Framework Programme to Project VALCRI under the EC Grant Agreement N° FP7IP-608142 awarded to Middlesex University and partners. Project Partners There are 17 project partners. Full details can be found at: http://valcri.org/partners-of-valcri/ Contact Details Project Coordinator, Professor William Wong Middlesex University London The Burroughs, Hendon London, NW4 4BT United Kingdom T: +44 (0)208 411 2684 E: w.wong@mdx.ac.uk W: www.valcri.org “Real world problems are multi-disciplinary. To cope with the variety of situations that may arise, real world solutions also need to be multi-disciplinary. In VALCRI, we have adopted a cognitive-behaviouraltechnological approach - cognitive engineering - where we combine technology in ways that enable humans to do what they are good at - reasoning and sense-making with ambiguity; and letting the machine do the heavy lifting, such as when given one report and to find others similar to it.”
Professor William Wong
William Wong is Professor of HumanComputer Interaction at Middlesex University. His main research interest is in the representation and design of information and the interaction of user interfaces to support decision making in complex dynamic environments.
www.euresearcher.com
75