Stable legal knowledge with regard to contradictory arguments
Shingo Hagiwara, Satoshi Tojo · International conference on Artificial intelligence and applications · 2006
A large size of legal knowledge base which consists of entangled inference rules, facts, and arbitrary interpretations may latently include inconsistency within them. In this paper, we propose a method to find the source of such inconsistency by supplying hypothesized facts into a set of rules. With this, we put those rules in the order of reliability and show a stable part of the legal knowledge. First, we define an argument as a chaining of rules to support a certain proposition. Thereafter, we compose a minimal inconsistency set (MIS) combining two disagreeing arguments. Among such a MIS, we can distinguish stable rules that is indifferent to the source of inconsistent from unstable rules, which can be candidates of future amendment. A knowledge-base which consists of stable rules can be also distinguished from that which may contain unstable rules.