In this paper we propose a generic task ontology, which formalizes the space of planning problems. Although planning is one of the oldest researched areas in Artificial Intelligence and attempts have been made in the past at developing task ontologies for planning, these formalizations suffer from serious limitations: they do not exhibit the required level of formalization and precision and they usually fail to include some of the key concepts required for specifying planning problems. In contrast with earlier proposals, our task ontology formalizes the nature of the planning task independently of any planning paradigm, specific domains, or applications and provides a fine-grained, precise and comprehensive characterization of the space of planning problems. Finally, in addition to producing a formal specification we have also operationalized the ontology into a set of executable definitions, which provide a concrete reusable resource for knowledge acquisition and system development in planning applications.

Watch the webcast replay >>