Split up: an intelligent decision support system which provides advice upon property division following divorce

In this paper we discuss Split Up, a hybrid rule-based/neural network system which provides advice upon the distribution of property following divorce in Australia. The task of determining what property a Family Court judge may distribute was determined to be rule-based and implemented using directed graphs. A hierarchy of ninety four factors relevant for a percentage split prediction were identified with the help of domain experts. The way some factors combine was later learnt by machine learning algorithms known as neural networks. The way other factors combined was modelled with rules that were derived from expert heuristics. Neural networks provide no explanation for their answers. The Split Up system utilizes a knowledge representation that is based on the argumentation theory of Toulmin. This enables the system to generate an explanation for conclusions reached. The system has been trialed by judges, judicial registrars and registrars of