Temporal constraint satisfaction on causal models

Abstract Causal models play a fundamental role in many intelligent problem solving activities such as diagnosis or simulation. Although the notion of causation is strictly related to the notion of time, temporal knowledge and temporal reasoning have been often disregarded in problem solvers operating on causal models. In this page, we propose an approach to temporal constraint satisfaction on causal models. The approach departs significantly from “traditional” approaches to temporal constraint satisfaction. In our approach, in fact, we use synchronization mechanisms for dealing with events caused by a conjunction of other events. In the paper, we analyze the computation complexity of temporal constraint satisfaction on causal models showing that such a form of reasoning is intrinsically intractable (i.e., it is intractable even if one limits the attention to a simple form of causal model and of consistency constraints). However, we show that it is possible to define sufficient conditions for temporal consistency that can be checked in polynomial time. Interestingly, such conditions allow us to relate our approach to “traditional” approaches to temporal constraint satisfaction.