HANDLING UNCERTAIN INTERVAL RELATIONS

摘要

Representation and reasoning about temporal information is important in many areas of AI. In many situations there is a need to deal with temporal relations, however, information about temporal relations might not always be perfect. In this paper we propose one way to represent uncertain temporal relations between temporal points and intervals using a probabilistic approach. Our representation for uncertain temporal relation between two primitives includes probabilities for the basic temporal relations that can hold between these primitives. We further propose three reasoning operations: inversion, composition, and addition. After that, we show how to translate uncertain interval relations using Allen's representation into the representation using the relations between the endpoints of these intervals, which can simplify further reasoning in some practical situations.