Argument interpretation using Minimum Message Length

Abstract

We describe an argument interpretation mechanism which receives as input a segmented argument composed of Natural Language sentences, and employs the Minimum Message Length Principle to select an interpretation among candidate options. This principle enables our mechanism to cope with noisy input in terms of wording, beliefs and argument structure. The performance of our system was evaluated by distorting automatically generated arguments, and passing them to the system for interpretation. Our evaluation showed that in most cases, the interpretations produced by the system matched precisely or almost-precisely the representation of the original arguments.