親しみやすい感性対話エージェントの実現に向けた動的嗜好更新手法

抄録

Recently, we frequently interact with computers, such as information guides, entertainment robots, etc. In order to make those interactions smooth, low-stress, and enjoyable, computer agents should not only exchange messages but also have emotions and tastes. Many methods for making computer agents have proposed. However, these methods do not consider taste changes. Therefore, we propose an agent which updates tastes and emotions through dialogues based on Heider's balance theory. Also, this paper reported the experiments of conversation with agents. We confirmed the effectiveness for enhancing agent's friendliness by the results of semantic differencial method and factor analysis.