A computational framework for dialectical reasoning

Dialectics are important not only in law but in every domain where knowledge is not certain; that is, everywhere assumptions must be made. After a review of recent advances in computational dialectics and related fields, we present the framework of a system for constructing dialectical arguments from a rule-based representation of law. In this system, meta level reasoning serves to allow for multiple utilisations of the rules. At the object level, rules grouped in modules represent “ground” knowledge. At the meta level, modules contain meta level rules that query other modules, at the object level or at some meta level, for arguments. During the construction of arguments, meta level rules use a filtering mechanism that works like simple regular expressions. This mechanism selects lower level rules according to their contexts. The object rules of the system are marked with interpretative contexts to permit varying points of view while maintaining an isomorphic representation of knowledge. The rules can be preceded by explicit negation, and the presence of contradictory rules allows conflicting arguments to be built. Examples are given and a discussion of future work concludes the paper.