Plan recognition for intelligent interfaces

A domain-independent assessment of plan recognition, with respect to its use in enhancing user interfaces, is presented. Plan recognition is a research area in automatic reasoning and is a promising approach to engineering better interfaces. Plan recognition makes interfaces more intelligent and interactive by providing an intelligent assistant that supports such tasks as advice generation, task completion, context-sensitive responses and error detection and recovery. How such tasks in turn provide representation and reasoning constraints that must be satisfied in order for the plan recognizer to efficiently support them is described. It is also shown how interfaces are fundamentally limited by current plan and recognition approaches. These limitations can be used to direct current research towards creating a new generation of plan-recognition systems.<
>