2018 Volume 30 Issue 5 Pages 744-752
In recent years, non-task-oriented dialogue system which aims at natural chatting of computer with human has received much attention. In order to increase the satisfaction of users, it is effective to generate utterances which can attract users’ interests in addition to performing natural conversation. Therefore, in this paper, we propose a method to acquire tweets that represents “ARUARU NETA” (typical story for many people) from Twitter and use them as system’s utterances. “ARUARU NETA” is one of the techniques in entertainment inviting laugh by getting empathy of the audience. As a result of the evaluation experiment of one round trip conversation, it was confirmed that the utterances of the proposed method were the most preferable compared with the chat dialogue API and normal tweets.
