Using an Activity Model to Address Issues in Task-Oriented Dialogue Interaction over Extended Periods

Collaborative natural dialogue offers a powerful medium for interaction between humans and intelligent autonomous agents (Allen et al. 2001). This is particularly the case for complex systems operating in dynamic, real-time environments, containing multiple concurrent activities and events which may succeed, fail, become cancelled or revised, or otherwise warrant discussion. Human dialogue in such scenarios is highly collaborative (Clark 1996), with much negotiation between instructor and task-performer: dialogue is used to specify and clarify goals and tasks, to monitor progress, and to negotiate the joint solution of any problems. Further, an interface to a device operating in such conditions must be interruptible, context-dependent, and otherwise extremely amenable to multiple threads of conversation. Dealing with interaction over an extended period—we envisage dialogues over a period of minutes to hours—raises specific challenges for systems operating in complex, dynamic environments. This is particularly the case when a user is operating multiple devices, requiring attentionswitching between devices and interactions, and long gaps between interactions with a specific device. In such circumstances, some of the specific challenges for task-oriented dialogue systems are the following: