In Crime science
Anti-social behaviour recorded by police more than doubled early in the coronavirus pandemic in England and Wales. This was a stark contrast to the steep falls in most types of recorded crime. Why was ASB so different? Was it changes in 'traditional' ASB such as noisy neighbours, or was it ASB records of breaches of COVID-19 regulations? Further, why did police-recorded ASB find much larger early-pandemic increases than the Telephone Crime Survey for England and Wales? This study uses two approaches to address the issues. The first is a survey of police forces, via Freedom of Information requests, to determine whether COVID-regulation breaches were recorded as ASB. The second is natural language processing (NLP) used to interrogate the text details of police ASB records. We find police recording practice varied greatly between areas. We conclude that the early-pandemic increases in recorded ASB were primarily due to breaches of COVID regulations but around half of these also involved traditional forms of ASB. We also suggest that the study offers proof of concept that NLP may have significant general potential to exploit untapped police text records in ways that inform policing and crime policy.
Halford Eric, Dixon Anthony, Farrell Graham
Anti-social behaviour, Antisocial behavior, Artificial intelligence, COVID-19, Natural language processing, Policing