Extending causal sequences to make teleological distinctions

Abstract

Summary form only given. The causal knowledge of how a specific behavior of a device arises from the functions of its components is useful for tasks like diagnosis, redesign, and prediction. This knowledge is usually represented using causal sequences (CSs). Annotations on the transitions in a CS often refer to the functions of device components as causes for the state transitions. The CS representation does not support any teleological distinctions between the functions of the components which is useful in tasks like redesign. Two different types of function, 'to achieve' and 'to prevent,' have been proposed earlier. However, the way CSs have usually been interpreted for various tasks implicitly assumes that the component functions are of the 'to achieve' type only. The author formalizes the current interpretive semantics of CSs, that is the inferences supported by CSs, and then extends these interpretive semantics to enable CSs to represent and reason about behaviors that refer to the 'to prevent' type of component functions as well. >