An Intelligent Process-driven Knowledge Extraction Framework for Crime Analysis


In recent years, mathematical, statistical and computational science methods have found extensive applications in developing new procedures for crime investigation, prosecution and the enforcement of law. Computer-based methods have also become important tools for performing certain forensic functions. Computational Forensic (CF) is an emerging interdisciplinary research domain. It concerns the investigation of forensic problems using computational methods, with the primary goal of discovery and advancement of forensic knowledge

Embedded in the context of this new interdisciplinary research domain, the general objective of the project described in this proposal (seen as collaboration between crime analysis experts and computer scientists working on data mining) is the development of an intelligent process-driven framework for crime analysis. The two components (sub-projects) cover the two research aspects: realization of a formal framework for modeling crime analysis processes (forensic subproject) and the development of an intelligent framework integrating knowledge (processes) and forensic data (computational sub-project).

The forensic sub-project, through a formalization of inference structures that pertains to crime analysis, should devise a set of areas where computational models may help to make the computer more active in the process. At the same time, it must describe how to interpret the patterns identified by the intelligent framework and how to relate them to current knowledge about crime analysis and environmental criminology. The task of modelling crime analysis processes rises two critical questions: (A) What is the appropriate formalism used to represent these structures, and (B) What is the appropriate methodology to generate and manage the collection of modeled processes?

The development of the computational framework for crime data analysis raises a number of theoretical and practical issues for which the computational sub-project must find appropriate solutions. The most critical of these issues are (i) What is the formal theory on which the reasoning/deduction must be performed, by considering the impreciseness and vagueness nature of forensic data and/or of crime analysis processes; (ii) How to implement in practice the integration/involvement of domain knowledge in the whole process of data analysis, what is the representation of this kind of knowledge, what is the mechanism allowing a crime analysis process execution to design and fine-tune data mining algorithms; (iii) How to ensure the coherence of a knowledge extraction iterative approach and how to evaluate, based on domain constraints, the interestingness and significance of the discovered patterns. The decision to ground the computational framework for crime data analysis on the formal theory of fuzzy systems may bring the necessary answers.

Personnes et institutions

Principal applicant Co-applicant PhD. students
Prof. Kilian Stoffel
Information Management Institute
University of Neuchâtel
Prof. Olivier Ribaux
Forensic Science Institute
University of Lausanne
Assist. Fabrizio Albertetti
Information Management Institute
University of Neuchâtel

Assist. Lionel Grossrieder
Forensic Science Institute
University of Lausanne

Données administratives

  • Date début : 01.09.2011
  • Date fin : 31.08.2015
  • Montant : 442 274 CHF
  • Financement : Fonds National Suisse


[1] K. Stoffel, D. Han and P. Cotofrei, "Fuzzy Methods for Forensic Data Analysis", Integration Techniques in Computational Forensics (ITCF'2010), Proceedings of International Conference SoCPaR 2010, pp. 23-28.

[2] P. Cotofrei and K. Stoffel, Fuzzy Extended BPMN for Modelling Crime Analysis Processes, In Proceedings of First Int. Symposium on Data - Driven Process Discovery and Analysis, SIMPDA  2011, pp. 13-28 .

[3] F. Albertetti and K. Stoffel, From Police Reports to Data Marts: a Step Towards a Crime Analysis Framework, International Workshop on Computational Forensics (IWCF) 2012, Japan.