Distinguishing ground from nonground information in defeasible argumentation

Abstract

The problem of speeding up ínference has proved to be important in argumentative systems. When computing dialectical structures, several paths are searched (called argumentation lines), many of which will eventually prove useless. Morover, the performance of backward chainers degrades quickly as the size of the knowledge base increases. This is a major hindrance for argumentative systems, sínce backward chaining is applied at two different levels (arguments themselves resemble proof trees, and arguments are related to each other within a tree structure). This paper presents a novel approach for speeding up inference. We will work with arguments for nonground literaIs, on the basis of some results presented by Levy [2]. As a result, most computations involving dialectical trees can be performed independently from the current groulld facts in the system 's knowledge base. The definition of abstraet dialectical tree will be introduced, which will account for aIl nonground queries for a given literal.