Towards Extending Domain Representations

Abstract

Based on our experience in developing and employing a rich domain representation for proof planning we compare the knowledge representation of the proof planner ΩMEGA with PDDL and with some AI planners and explore the usefulness of some extensions for Al-planning more generally. In particular, we investigate the more expressive syntax and semantics of proof planning due to the introduction of functions, full quantification, integration of constraint solving, powerful filters, and context-dependency as well as knowledge implicit in integrated systems.