Adaptive mixed-initiative dialogue management

This paper describes an architecture that supports adaptive mixed-initiative dialogue. It is based on a generalisation of the form-filling paradigm. Rather than a flat slot structure, we use a hierarchy that contains slots at various levels of abstraction. Along with the slot hierarchy, a question hierarchy is defined that allows for adaptive mixed-initiative dialogue. Depending on the success or failure of certain questions, the system can zoom-in to more detailed questions, or zoom-out to higher-level questions. The distribution of initiative in dialogue is closely related to the granularity of the information that is asked for. To determine the right granularity level for system questions we have to take into account the influences on user freedom, predictability and the knowledge of the user. Giving the user initiative is suitable in situations where the user knows what to do. In such cases, the user may give all relevant information in one turn. However, giving the user more initiative tends to make his behaviour less predictable and therefore increases the chance on speech-understanding errors. The system should switch to lower-level questions when higher-level ones fail, and when the user supplies unsolicited information, the system can switch to higher-level questions.