When stories and numbers meet in court: Constructing and Explaining Bayesian Networks for Criminal Cases with Scenarios

When a judge or jury is presented with evidence in a criminal trial, they must apply some sort of reasoning process to draw a conclusion from this evidence. For instance, from a witness testimony they might conclude that the suspect was at the crime scene, and from fingerprints on a murder weapon they might conclude that the suspect used the weapon. With the rise of modern forensic techniques such as DNA profiling, such legal evidence often comes with some sort of probabilistic information. For example, when forensic experts report on a DNA match, they will typically report a so-called random match probability: the probability that the match would be found if the suspect were not the source of the DNA trace that was tested. A challenging task for a judge or jury is to take into account the whole case, while incorporating probabilistic information on elements of the case. In this thesis we developed a method in which probabilistic evidence can be modelled in the context of scenarios. Scenarios provide an intuitive way for a judge or jury to make sense of the case as a whole. Using so-called Bayesian networks as an existing modelling technique, in our method scenarios are used to construct a probabilistic model of the case and scenarios can also be used to explain this model to a judge or jury. Our proposed method lays the foundations for a software support system, for instance in the form of a computer programme to support a judge or jury in their reasoning process