Computing Ramifications by Postprocessing

Abstract

A solution to the ramification problem caused by underlying domain constraints in Stripslike approaches is presented. We introduce the notion of causal relationships which are used in a post-processing step after having applied an action description. Moreover, we show how the information needed for these post-computations can be automatically extracted from the domain constraints plus general knowledge of which fluents can possibly affect each other. We illustrate the necessity of causal relationships by an example that shows the limitedness of a common method to avoid unintended ramifications, namely, the distinction between so-called frame and non-frame fluents. Finally, we integrate our solution into a recently developed, Strips-like yet purely deductive approach to reasoning about actions based on Equational Logic Programming. 1 Introduction The ramification problem [ Finger, 1987 ] is usually regarded as one of the challenges to all formal frameworks for reasoning about actions a...