AQMeN Seminar Series

Description

'Criminal Evidence Evaluation using Graphs' Abstract: The use of graphs to illustrate the structure of a criminal investigation is not new, being first introduced by Wigmore in 1937 with tree diagrams. A more recent probabilistic approach to the evaluation of evidence using graphs will be described. Pieces of evidence and propositions are represented by nodes and associations between pieces of evidence and propositions are represented by lines joining nodes. The strengths of the associations are measured by probabilities. The technique will be illustrated from an example in offender profiling motivated by an investigation of sexually motivated child murders in the 1990’s and, more recently, the case of R v T (2010, EWCA Crim 2439). A graph that describes a relationship between evidence and a proposition would consist of two nodes joined by a directed edge (arrow) pointing from the node representing the proposition to the node representing the evidence. The proposition can take one of two values: for example, a pair of propositions could be that a defendant is guilty or is innocent, another pair could be that the defendant was or was not the source of the DNA found at the crime scene. The directed edge illustrates the relationship of evidence with this proposition and that there is information about the probability of the evidence if the defendant is guilty (or was the source of the DNA at the crime scene) and the probability of the evidence if the defendant is innocent (or was not the source of the DNA at the crime scene). Rules of probability can be used to reverse the inference to derive probabilities for the propositions given the evidence and hence evidential value. The idea can be applied to graphs with many nodes, representing different pieces of evidence and different propositions, and many edges representing relationships amongst the pieces of evidence and the propositions.

Period	26 Apr 2012
Event type	Seminar
Location	Edinburgh, United KingdomShow on map