Abstract

Recently, a great deal of progress has been made using nonmonotonic temporal logics to formalize reasoning about action and change. In particular, much focus has been placed on the proper representation of non-deterministic actions and the indirect effects of actions. One popular approach to representing the indirect effects of actions has been via the use of causal rules which in a more general sense can be viewed as fluent dependency constraints. Although fluent dependency constraints have been used primarily under a loose causal interpretation, we show that when interpreted in a broader sense they provide a flexible means for dealing with a number of other representational problems such as the qualification problem and the ramification constraints as qualification constraints problem, in addition to the standard ramification problem. More importantly, the use of fluent dependency constraints for different purposes does not involve additions to the base nonmonotonic temporal logic, TAL, used here, but simply the addition of several macro operators to an action language used to represent action scenarios or narratives. The payoff is that TAL has already been shown to offer a robust approach to representing action scenarios which permit incomplete specifications of both state and the timing of actions, non-deterministic actions, actions with duration, concurrent actions, use of both boolean and non-boolean fluents, and solutions to the frame and ramification problems for a wide class of action scenarios. In addition, all circumscribed action scenarios in these classes and the more general class involving qualification considered in this paper can be shown to be reducible to the first-order case. Finally, a restricted entailment method for this new class of scenarios is fully implemented. In the paper, we present a challenge example which incorporates all these features, propose a distinction between weak and strong qualification with a representation of both, and provide a visualization of the preferred entailments using a research tool VITAL for querying and visualizing action scenarios.