1993 Reports

Terminological Constraint Network Reasoning and its Application to Plan Recognition

Weida, Robert Anthony

Terminological systems in the tradition of KL-ONE are widely used in AI to represent and reason with concept descriptions. They compute subsumption relations between concepts and automatically classify concepts into a taxonomy having well-founded semantics. Each concept in the taxonomy describes a set of possible instances which are a superset of those described by its descendants. One limitation of current systems is their inability to handle complex compositions of concepts, such as constraint networks where each node is described by an associated concept. For example, plans are often represented (in part) as collections of actions related by a rich variety of temporal and other constraints. The T-REX system integrates terminological reasoning with constraint network reasoning to classify such plans, producing a "terminological" plan library. T-REX also introduces a new theory of plan recognition as a deductive process which dynamically partitions the plan library by modalities, e.g., necessary, possible and impossible, while observations are made. Plan recognition is guided by the plan library's terminological nature. Varying assumptions about the accuracy and monotonicity of the observations are addressed. Although this work focuses on temporal constraint networks used to represent plans, terminological systems can be extended to encompass constraint networks in other domains as well.