Modeling user interpersonal stances in affective dialogues with an ECA

Since several forms of anthropomorphic behavior of users towards technologies have been demonstrated, ongoing research on intelligent interfaces aims at developing adapting conversational systems which tailor their behavior according to both cognitive and affective features of users. We present an approach for combining results of our previous experience in detecting and modeling interpersonal stances in advice-giving dialogues with an ECA. We sketch an algorithm for adaptation of the behavior of the agent according to the user social attitude and level of engagement