Transaction Logic with Defaults and Argumentation Theories

Abstract

Transaction Logic is an extension of classical logic that gracefully integrates both declarative and procedural knowledge and has proved itself as a powerful formalism for many advanced applications, including modeling robot movements, actions specification, and planning in artificial intelligence. In a parallel development, much work has been devoted to various theories of defeasible reasoning. In this paper, we unify these two streams of research and develop Transaction Logic with Defaults and Argumentation Theories, an extension of both Transaction Logic and the recently proposed unifying framework for defeasible reasoning called Logic Programs with Defaults and Argumentation Theories. We show that this combination has a number of interesting applications, including specification of defaults in action theories and heuristics for directed search in artificial intelligence planning problems. We also demonstrate the usefulness of the approach by experimenting with a prototype of the logic and showing how heuristics expressed as defeasible actions can significantly reduce the search space as well as execution time and space requirements.