Reasoning about unpredicted change and explicit time
Florence Dupin de Saint-Cyr, Jérôme Lang
TL;DR
Reasoning about unpredicted change with explicit time addresses explaining time-stamped observations via surprises, i.e., unpredicted changes in fluents not caused by agent actions. The authors present a formal framework, define minimal and compact explanations (Cme(Σ)) within the K-IS class, and show how to compute them using persistence axioms and ATMS-inspired mechanisms, aligning with model-based diagnosis. They extend the framework with a probabilistic layer by modeling fluents as Markovian and deriving posterior probabilities that favor highly persistent fluents and shorter change intervals, enabling ranking of explanations. The approach yields concise, temporally grounded explanations and offers a path to extending to dependent fluents and action-based scenarios, bridging temporal diagnosis and probabilistic reasoning.
Abstract
Reasoning about unpredicted change consists in explaining observations by events; we propose here an approach for explaining time-stamped observations by surprises, which are simple events consisting in the change of the truth value of a fluent. A framework for dealing with surprises is defined. Minimal sets of surprises are provided together with time intervals where each surprise has occurred, and they are characterized from a model-based diagnosis point of view. Then, a probabilistic approach of surprise minimisation is proposed.