Study on Multi-modal Persuasive Agents

我们研究了能够以语言和非语言方式说服用户的仿生代理技术，并开发了应用系统。作为一种言语说服技术，我们利用《绿野仙踪》的方法开发了学习说服策略的代理。作为一种非语言说服技术，我们研究了反映用户话语速度的聊天系统如何影响用户的印象。作为一种多智能体说服技术，我们研究了多智能体和用户之间的社会关系对说服效果的影响。特别是我们发现增强群体认同对提高绩效具有重要意义。我们开发了一个说服性购物网站作为我们的说服性技术的应用系统。

We investigated technologies for life-like agents which can persuade users in verbal and non-verbal ways and developed application systems. As a verbal persuasive technology, we developed agents which learn persuasion strategies by using the Wizard of Oz method. As a non-verbal persuasive technology, we investigated how a chat system, which reflects the utterance speed of the user, affects its user's impression. As a multiagent persuasive technology, we investigated how social relationships between multiple agents and users affect the performance of persuasion. Especially we found it is significant to enhance the group identification to improve the performance. We developed a persuasive shopping site as an application system of our persuasive technologies.