Split Up: The use of an argument based knowledge representation to meet expectations of different users for discretionary decision making

摘要

Split Up is a rule / neural hybrid that respresents knowledge using frames based on the argument structure proposed by the British philosopher, Toulmin. Split Up makes predictions abut marital property following a divorce in Australia; a domain that is considered sicretionary in that a judge has considerable flexibility. The end users of Split Up are judges and registrars of the Family Court of Australia, mediators and lawyer. Each end user has specific and divergent needs and thus uses the system in different ways however all users rely on effective explanations. The argument based representation of knowledge enables the system to have the flexibility required of different users, to generate effective explanations and als facilitates knowledge acquisition. The framework has been used to integrate rules with neural networks but can easily be used to integrate other inferencing methods.