Resumen de Proclaim: an argument-based model for deliberating over safety critical actions
Francisco Tolchinsky Landsmann
In this Thesis we present an argument-based model ProCLAIM intended to provide a setting for heterogeneous agents to deliberate on whether a proposed action is safe. That is, whether or not a proposed action is expected to cause some undesirable side effect that will justify not to undertake the proposed action. This is particularly relevant in safety- critical environments where the consequences ensuing from an inappropriate action may be catastrophic.
For the practical realisation of the deliberations the model features a mediator agent with three main tasks: 1) guide the participating agents in what their valid argumentation moves are at each stage of the deliberation; 2) decide whether submitted arguments should be accepted on the basis of their relevance; and finally, 3) evaluate the accepted arguments in order to provide an assessment on whether the proposed action should or should not be undertaken, where the argument evaluation is based on domain consented knowledge (e.g guidelines and regulations), evidence and the decision maker expertise. To motivate ProCLAIM's practical value and generality the model is applied in two scenarios: human organ transplantation and industrial wastewater. In the former scenario, ProCLAIM is used to facilitate the deliberation between two medical doctors on whether an available organ for transplantation is or is not suitable for a particular potential recipient (i.e. whether it is safe to transplant the organ). In the later scenario, a number of agents deliberate on whether an industrial discharge is environmentally safe.
