Modeling crime scenarios in a Bayesian Network

Legal cases involve reasoning with evidence and with the development of a software support tool in mind, a formal foundation for evidential reasoning is required. Three approaches to evidential reasoning have been prominent in the literature: argumentation, narrative and probabilistic reasoning. In this paper a combination of the latter two is proposed.

In recent research on Bayesian networks applied to legal cases, a number of legal idioms have been developed as recurring structures in legal Bayesian networks. A Bayesian network quantifies how various variables in a case interact. In the narrative approach, scenarios provide a context for the evidence in a case. A method that integrates the quantitative, numerical techniques of Bayesian networks with the qualitative, holistic approach of scenarios is lacking.

In this paper, a method is proposed for modeling several scenarios in a single Bayesian network. The method is tested by doing a case study. Two new idioms are introduced: the scenario idiom and the merged scenarios idiom. The resulting network is meant to assist a judge or jury, helping to maintain a good overview of the interactions between relevant variables in a case and preventing tunnel vision by comparing various scenarios.