Exploring Applications of AI for Monitoring & Detecting Changing Crime Problems

The nature and scope of crime is changing with the advent of new technologies that offer (1) new methods for committing existing crimes; (2) new opportunities for committing new types of crime; and (3) new means to prevent and detect crime. These changes present both significant challenges and opportunities to police and their crime reduction partners. Consequently, there are significant benefits to be gained from developing methods capable of identifying the changing nature of crime problems. In response, this project will explore the application of AI techniques for detecting new crime problems in recorded crime data that is routinely collected by policing agencies but seldom analysed - with the aim of informing the development of early warning systems capable of targeting more effective service delivery.

Each year policing agencies and their partners collect increasingly large volumes of administrative data primarily for operational and housekeeping purposes. For a number of reasons these data are vastly underutilised in comparison to the collective investment in their capture. To illustrate, police routinely record ‘modus operandi’ free-text data describing the means by which an offence was committed. The large volume and unstructured nature of these data dictate that they cannot be analysed en masse using existing analytical approaches: instead they are used only for investigatory purposes on a case-by-case basis. Moreover, the nature of police crime recording systems dictate that emerging trends in offending cannot be systematically identified without considerable work of a police analyst. Consequently, without means to automate strategic analyses of these data, important trends and patterns can be missed. Three recent examples of emerging problems for police not easily captured by traditional crime analyses include burglaries targeting the keys of luxury vehicles to enable their subsequent theft, ride-by mobile phone thefts by pairs of offenders on mopeds, and the use of security bypass technologies and RFID key code interception techniques employed by technologically sophisticated vehicle thieves. In an attempt to bridge this significant analytical gap, this project will explore the effectiveness of several AI techniques in deriving actionable insights from largely untapped sources of police data.

To achieve this goal the project will apply AI methods to analyse a range of unstructured and semi-structured police recorded crime data with the aim of detecting new crime problems as they emerge. It will begin by analysing large quantities of historic crime data describing modus operandi and property targeted. These data will be used to explore, develop and test a series of approaches for automating the detection of changes in types/methods of crimes being committed. Subsequently, combining these analyses with traditional recorded crime data (location, time-date, etc.) a typology of crime opportunity trajectories and their diffusion characteristics will be constructed. Finally, these insights will be combined to inform the development of prototype early warning systems capable of passively monitoring crime recording systems to detect changes in the way crimes are committed and the types of targets they are committed against – with the aim of highlighting areas for early-targeted intervention.

Research Questions The project will address the following overarching research questions:

Can AI methods be used to identify changing methods of, and opportunities for, committing crime detailed in unstructured and semi-structured recorded crime data?

Can these techniques be incorporated into early warning systems in order to optimise the targeting of service delivery (further analytics, crime prevention efforts, police operations, etc.)?

Entry requirements

Minimum UK Upper Second Class Honours AND Merit at Masters or equivalent. English Language requirements are higher than University minimum. Background in computer science (BSc/MSc) or related fields, preferably with experience of machine learning.

How to apply

How to Apply for an ESRC WRDTP Studentship at Leeds:

  1. Applicants applying on both a +3 and 1+3 basis should first of all apply for the relevant research postgraduate programme to commence 1 October 2018 and obtain their Student ID Number;
  2. Applicants should complete the research on-line scholarship application form available at: https://leeds.onlinesurveys.ac.uk/esrc-white-rose-dtp-studentships-university-of-leeds-ai by the relevant deadline.  An Information Sheet with regulations, guidance notes and a link to the application form is available at http://scholarships.leeds.ac.uk/Documents/esrc-2744.pdf

It is important that you select the correct studentship type for which you wish to apply, as this will determine the sections you will be asked to complete.

After receipt of your studentship application, the relevant School will provide further advice on your suitability for either a 1+3 or +3 studentship and advise whether you are required to apply for a Masters study place for the MA Social Research (Interdisciplinary) programme

Applicants on a +3 basis would be expected to commence PhD study by no later than January 2019.

Applicants on a 1+3 basis would be expected to commence the Masters programme in September 2019.

How to apply (email)


How to apply (phone)

+44 113 343 5009