Recently, preference information is
often utilized in the field of information
recommendation, human-computer interaction,
and so on. We previously proposed a method to
learn object’s preference from grammatical
feature of utterance. However, it is difficult to
collect such type of utterance enough because the
method requires strict structure of grammar. In
this paper, we suggest a new method to learn
object’s preference by calculating speaker’s
emotion based on acoustic features.

開催日：平成24年7月14日　会場：広島市立大学