Dialogue Generation in an Assistive Conversation Skills Training System for Caregivers of Persons with Alzheimer's Disease

We describe a proposed assistive conversational skills training system using artificial intelligence and natural language generation techniques to simulate spoken dialogue between two embodied agents, one character representing a caregiver and the other character a person with Alzheimer's Disease (AD). The type of dialogue simulated is social conversation in which the caregiver encourages the person with AD to tell autobiographical stories. Intervention by the caregiver may be required at times to keep the conversation going due to linguistic or cognitive problems experienced by the storyteller. In our proposal, turn and topic management behavior for the storyteller character is implemented by combining rules representing normal pragmatic routines with linguisticallymotivated rules representing coping strategies.