感性情報の学習と収集を行う対話型デスクトップマスコットの提案

概要

論文の詳細を見る
In this paper, we propose a new desktop mascot model for entertainment and information gathering. Unlike most conventional desktop mascots, proposed agent can learn words from its user, as well as emotions attached to those words. Based on past learning, the agent shows different emotionsto us with different topics. For example, if the user makes positive feedbacks to topics about a particular company, the agent is likely to smile when the same company is mentioned again. Therefore, this agent can provide more interactive, dynamic and fun interaction for the user.The other functionality of this agent is to accumulate a database for its supervisor.Through the interaction with users, the agent can leam new proper nouns, technical terms, jargons and slangs.In addition, experiments we carried out showed that the proposed model effectively gathered users, emotional feelings toward those words.