Toby Donaldson and Robin Cohen
In this paper, we discuss how a discourse agent can manage multiple discourse goals using dynamic constraint optimization. We discuss turn-taking agents, which are agents that know how to take turns in dynamic, conversational settings. Conversation constrains an agent’s behaviour in a number of ways; for example, a conversational agent must act quickly to avoid long pauses, and it must be prepared to sometimes act before it is completely finished thinking. Our solution to this turn-taking problem satisfies these conattaints, and also provides for straightforward application of well-known heuristics for solving the particular optimization problem that arises.