1st International ICST Workshop on Computational Forensics

Research Article

Application of Language Models to Suspect Prioritisation and Suspect Likelihood in Serial Crimes

  • @INPROCEEDINGS{10.1109/IAS.2007.58,
        author={Richard Bache and Fabio  Crestani and David Canter and Donna Youngs},
        title={Application of Language Models to Suspect Prioritisation and Suspect Likelihood in Serial Crimes},
        proceedings={1st International ICST Workshop on Computational Forensics},
        publisher={IEEE},
        proceedings_a={IWCF},
        year={2007},
        month={9},
        keywords={Computer crime  Data analysis  Data security  Databases  Frequency  Information analysis  Information retrieval  Information science  Information security  Psychology},
        doi={10.1109/IAS.2007.58}
    }
    
  • Richard Bache
    Fabio Crestani
    David Canter
    Donna Youngs
    Year: 2007
    Application of Language Models to Suspect Prioritisation and Suspect Likelihood in Serial Crimes
    IWCF
    IEEE
    DOI: 10.1109/IAS.2007.58
Richard Bache1,*, Fabio Crestani1,*, David Canter2,*, Donna Youngs2
  • 1: Department of Computing and Information Science, University of Strathclyde, Scotland
  • 2: Centre for Investigative Psychology, England
*Contact email: rb@cis.strath.ac.uk, fabioc@cis.strath.ac.uk, dcanter@ukonline.co.uk

Abstract

Language models are successfully applied to the problem of analysing crime descriptions from a police database with the purpose of prioritising suspects for an unsolved crime, given details of solved crimes. The frequency of terms in each description relates to the behaviour of the offender and this can be used to link crimes to a common offender. Language modelling uses Bayes' theorem and thus require a prior probability. Such a prior can be based on each offender's past propensity to offend, derived from historic data. Language modelling yields a probability of a document being relevant, which in this case is interpreted as the probability of a suspect being the culprit. Although the absolute value of the probability does not carry any direct applied implications, the study does show that the general likelihood of identification of the actual suspect does correspond to the relative values. Thus these probabilities can be used for more than just ranking suspects.