Jusletter IT

Legal Reasoning involves both analytic/deductive and ampliative/analogical components. The latter are tightly connected to defeasibility in legal argumentation. Yet present rule-based models of defeasible legal inference are not ampliative: what is defeasibly derivable is solely determined by the contents of their knowledge base. We argue that a proper modelling of the analogical aspects of legal reasoning allows one to capture both much of its ampliative and much of its defeasible capabilities. Context graphs enable the representation of ampliative analogies as metalogical operations. We sketch algorithms acting on Context Graphs to implement analogical reasoning.