Pertinence for causal representation of action domains
Pedro Cabalar · 2001
This dissertation goes deeply into the use of causality for reasoning about actions and change, focusing not only in its application for solving typical representational problems, but also on causal knowledge as a relevant information per se. We show that causality is intimately related to the concept of pertinence with respect to the actions execution, and we propose a formal characterization of this idea, explaining its role in the semantics of causal rules. This allows us to introduce a high level language that provides a comfortable representation of action domains by using causal constructions. Besides, we present a uni ed view of the two existing orientations about causality, incorporating them at two dierent levels: at an ontological level, we use pertinence to capture the idea of causality versus persistence; whereas at an epistemological level, we propose dierent non-monotonic techniques to achieve the directional behavior of causal expressions.