Using SPARQL to express Causality in Explainable Cyber-Physical Systems

Causality knowledge is an essential component of an explainable cyber-physical systems. Since this knowledge mainly come from domain experts, manual annotation of causality is prone to errors especially if the system is large and complex. This paper describes how SPARQL queries can minimize domain expert work by (1)partitioning causality as two level of abstraction: abstract and concrete causality, and (2)inferring concrete causality from query result. We illustrate the usefulness of this method by showing the use of SPARQL in a smart grid scenario.