XAI-LAW Towards a logic programming tool for taking and explaining legal decisions

In this paper we present an overview of a research project aiming at producing a semi-automated tool for legal reasoning in the Italian criminal system during the criminal trial stage. Parts of the Italian Criminal Law are modeled in ASP; the model obtained is tested on a set of previous statements on the crimes, and, if needed, refined. Decisions on a new case can be suggested by the system and explained using a tool that exploits “supportedness” of stable models. In the same way, the decision of a judge can be input in the system and automatically explained. Using a system of inductive logic programming for ASP, the tool can evolve by analyzing new statements and performing model revision, by learning exceptions, and by applying rule generalization. To study feasibility of the approach we analyzed the crimes of theft, robbery, and personal injuries. Further crimes will be considered in the future development of the project.