Chapter

Innovations in Applied Artificial Intelligence

Volume 3029 of the series Lecture Notes in Computer Science pp 689-698

Potential Causality in Mixed Initiative Planning

  • Yousri El FattahAffiliated withRockwell Scientific

* Final gross prices may vary according to local VAT.

Get Access

Abstract

The paper presents an approach for reasoning about potential causality in plans authored directly by humans in a mixed initiative framework. The approach uses only the temporal ordering of the actions and the task structure of the plan. The term potential is used to emphasize the uncertainty in the causal ordering since no requirement is made on the existence of a complete domain theory as in standard partial order planning. The core contribution of the paper is a formalization and algorithm for extracting a parsimonious description of a potential causality relation, which is presented to the modeler as a representation of the candidate space of sets of causal links consistent with the authored plan. The paper also discusses an implemented system based on this algorithm, and its application in the context of execution.