Building an Intelligent Crime Linkage System
Software engineering is a branch of engineering that focuses on understanding the processes and methods that are utilised in the design, development and maintenance of software. The philosophy is that research in this area will create methods that developers can use to ensure they write robust software that meets user requirements, is designed for the correct environments and accounts for system needs. For Dr Dalal Alrajeh a Junior Research Fellow at Imperial, her focus is on developing tool-supported approaches to aid software engineers using techniques such as logic and reasoning. In recent years, these interests have extended beyond traditional software engineering to explore how these techniques can be applied to other areas where robust understanding of systems and data can provide support and insight to users of that data.
In 2011, Dalal was presenting at a British Computer Society workshop in East Anglia where by chance one of the organisers worked for the police service in IT who started to explore with Dalal how software engineering approaches might be used by the police force to improve the services they provide. Over the course of time, the collaboration has developed such that Dalal is now supported by the ESRC IAA to collaborate with the Serious Crime Analysis Section (SCAS) of the National Crime Agency (NCA) to develop intelligent crime linkage support tools.
The 2013/2014 Crime Survey for England and Wales estimated that there were 1.3 million incidents of violence against adults and 445,000 violent offences against children in England and Wales in the previous 12 months. The number of police recorded sexual offences in the year to March 2014 showed a 20% increase compared with the previous year, rising to a total of 64,205 incidents across England and Wales. This latest figure is the highest ever recorded on a financial year basis. Within this, the number of offences of rape increased by 26% to 20,745 incidents, and the number of other sexual offences increased by 17% to 43,460 incidents. Methodologies to support the police identify which of these crimes are committed by the same perpetrator is crucial for the development of efficient joint investigative strategies within and between forces.
Crime linkage is a technique used by UK police forces to identify common behaviours in crimes (e.g. perpetrator’s method of attack, escape, controlling victims between recorded crimes). By identifying these commonalities, the police are able to establish and understand incidences of serious and violent serial crimes. Analysts working in this field will typically have many years’ experience working as an analyst in which they have developed a unique skill set and insight into their work. However, as with any work that requires human analysis of complex sets of data, analysis may be subjective and a one analyst can offer different insights and perspectives to others. As the volume of cases processed by analysts is high, and pressure to deliver results ever-increasing, the cognitive load escalates. This may lead to discrepancies in the way linkage methods are applied and results are interpreted, with inevitable implications on the wider investigative process.
Typically an analyst may work on a case for several weeks, meaning that the time for multiple analysts to devote time to the same case is often stretched. For Dalal, working with the NCA provided an opportunity to look at developing a support tool that could potentially benefit analysts by providing insights from the data about potential crime linkages. However, to achieve this there would be several obstacles to overcome.
Over the past few months Dalal has been working with some of the NCAs to understand what are the triggers/indicators that an analyst is identifying when they are picking out commonalities and linkages between individual cases in order to build intelligent systems that can support these decision making processes. This task in itself is hugely difficult as analysts are often having to try to articulate thought processes that are abstract and complex. From these descriptions Dalal has to create and formalise a series of rules using her logic and reasoning techniques that the support tool will then be able to use to identify and explain possible linkages in the recorded crime data.
Dalal hopes that the tool will eventually provide analysts with a system that can help them refine their thinking and analysis by effectively providing access to the logic and experience of several different analysts’ perspectives on potential crime linkages. To date she has created a prototype for the tool which will go through further refinement and development stages later in the autumn before being showcased to a range of crime analysts in the new year.