From traces to knowledge: Toward computational forensic criminology ?

L.Grossrieder1*, F. Albertetti2, K. Stoffel2, O. Ribaux1

1 School of Criminal Justice, University of Lausanne
2 Information Management Institute, University of Neuchâtel

*corresponding author:

The role of statistics and computational models in crime analysis is hardly contestable. The concrete added value provided by this global movement is however far from obvious and proposed approaches show many limitations, occasionally proved to be unrealistic. This presentation focuses on the contribution of computational techniques in crime analysis, especially for crime trends detection, through the lens of an interdisciplinary framework for better situating and integrating these innovations.

The proposed approach is based on a fundamental postulate of crime analysis: crimes follow patterns that can be detected and analysed through the exploitation of accessible data. The arguments are founded on the most elementary piece of data available: the trace, physical (and numerical) remnant of the litigious activity, which has been recognized and collected at crime scenes. To illustrate this framework, we focus on crime trends detection in order to explore the possibility to detect automatically these patterns or these inconsistencies with a change point analysis.

To empirically reach this objective, we have analysed police data of the canton of Vaud in Switzerland composed by serial or itinerant crime events and crime trends detection identified by crime analysts. The results show that automatic detection of breaks relating to crime trends is possible and strongly accurate with suitable parameters. Moreover, they show the necessity to build a more ambitious interdisciplinary framework in crime science, which will help to structure further the approach. Called Computational Forensic Criminology (CFC), it will seek to deliver crime analysis and intelligence, with the means of crime data stemming from traces, analysed with computational methods, and explained/supported by criminological theories.

This communication is part of the Swiss National Science Foundation project “An Intelligent Process-driven Knowledge Extraction Framework for Crime Analysis”, supported by the grant no156287 of the Swiss National Science Foundation.


Lionel Grossrieder is a research associate and a doctoral student in the School of Criminal Justice of Lausanne, Switzerland. He received his bachelor degree in psychology in 2009 and his master degree in criminology in 2011. His research interests include crime analysis, environmental criminology and forensic intelligence. He is currently involved in an interdisciplinary project on application of computational methods in crime analysis.