Research on crime and security issues is experiencing a push towards data science fuelled by new challenges and stakeholder demand. With an increasing interest in understanding problems at scale, automated methods and data-driven approaches are embraced with the promise of solving long-standing and emerging problems. The first part of the talk presents recent research that shows the potential and the pitfalls of the rapid adoption of data science techniques in crime research. Two studies are presented that cover the modelling of the temporal evolution of a white nationalist forum using trajectory analysis, and a large-scale measurement of an online grey-zone forum on the sharing of non-sexually explicit imagery of children. The second part of the talk discusses the future of data science in a broader sense in the tension between applications and academic research. It addresses the question of why we need to embrace data science but also why we should be cautious about doing so. The talk will close with an outlook on the challenges for the merger of the behavioural and social sciences with data science.
About:
Dr Bennett Kleinberg is an assistant professor in Data Science at University College London. In the Dept. of Security and Crime Science and the Dawes Centre for Future Crime, he is leading a research group on computational crime research that has two aims. (1) Understanding how computational methods can be used to study current and emerging crime problems (esp. through technological advancement); and (2) building a robust methodological framework for behavioural data science to improve how we make inferences about human behaviour from complex data (esp. text and web data). These areas include techniques from natural language processing, machine learning and behavioural science.