A Legal Reasoning System with Event and Property Interpretation for Legal Knowledge

摘要

In legal reasoning, the knowledge representation is one of the toughest problems, and thus far many subtle representation of natural language have been lost when they have been formalized. In particular, when the legal knowledge is represented in the form of predicate logic, the distinction between one occurrence of an event and a universal property has been neglected and thus the implicit meaning of the original rules (and therefore the ability to make inferences from them) has been lost. In this paper, we distinguish events and proper- ties in predicates, as the problem of quantification. We introduce a legal reasoning system based on order-sorted logic, where we hierarchically sort both of predicates and objects, and distinguish events from properties by superordinate-predicate derivation and quantification of implicit objects. As a result, we would be able to apply legal rules to flexible inference that operate on both properties and events. We implement a logic programming system with unification/resolution mechanisms, where the distinction is realized as the substitution of sort and the manipulation of arguments, and show the feasibility.