Cluster-based approach to discriminate the user’s state whether a user is embarrassed or thinking to an answer to a prompt

Human–machine metacommunication towards development of a human-like agent: A short review

“…This kind of ability to estimate the metacognitive state of the interlocutor is indispensable for a dialog system to realize a human-like conversational agent. There are only a few research works in this field, for example, Chiba et al attempted to discriminate what kind of difficulty the speaker has when the speaker does not give a response [29]. Recognizing the awareness and understanding of a speaker is especially important in multiparty conversation [30].…”