Predicting and Eliciting Addressee’s Emotion in Online Dialogue

Proceedings of the 51st Annual Meeting of the Association for Computational Linguistics, pages 964–972, Sofia, Bulgaria, August 4-9 2013. c⃝2013 Association for Computational Linguistics Predicting and Eliciting Addressee’s Emotion in Online Dialogue Takayuki Hasegawa∗ GREE Inc. Minato-ku, Tokyo 106-6101, Japan takayuki.hasegawa@gree.net Nobuhiro Kaji and Naoki Yoshinaga Institute of Industrial Science, the University of Tokyo Meguro-ku, Tokyo 153-8505, Japan {kaji,ynaga}@tkl.iis.u-tokyo.ac.jp Masashi Toyoda Institute of Industrial Science, the University of Tokyo Meguro-ku, Tokyo 153-8505, Japan toyoda@tkl.iis.u-tokyo.ac.jp Abstract While there have been many attempts to estimate the emotion of an addresser from her/his utterance, few studies have ex- plored how her/his utterance affects the emotion of the addressee. This has mo- tivated us to investigate two novel tasks: predicting the emotion of the addressee and generating a response that elicits a specific emotion in the addressee’s mind. We target Japanese Twitter posts as a source of dialogue data and automatically build training data for learning the pre- dictors and generators. The feasibility of our approaches is assessed by using 1099 utterance-response pairs that are built by five human workers. 1 Introduction When we have a conversation, we usually care about the emotion of the person to whom we speak. For example, we try to cheer her/him up if we find out s/he feels down, or we avoid saying things that would trouble her/him. To date, the modeling of emotion in a dialogue has extensively been studied in NLP as well as re- lated areas (Forbes-Riley and Litman, 2004; Ayadi et al., 2011). However, the past attempts are vir- tually restricted to estimating the emotion of an addresser1 from her/his utterance. In contrast, few studies have explored how the emotion of the ad- dressee is affected by the utterance. We consider the insufficiency of such research to be fatal for ∗This work was conducted while the first author was a graduate student at the University of Tokyo. 1We use the terms addresser/addressee rather than a speaker/listener, because we target not spoken but online di- alogue. I have had a high fever for 3 days. JOY I hope you feel better soon. I have had a high fever for 3 days. SADNESS Sorry, but you can’t join us today. Figure 1: Two example pairs of utterances and re- sponses. Those responses elicit certain emotions, JOY or SADNESS, in the addressee’s mind. The ad- dressee in this example refers to the left-hand user, who receives the response. computers to support human-human communica- tions or to provide a communicative man-machine interface. With this motivation in mind, the paper inves- tigates two novel tasks: (1) prediction of the ad- dressee’s emotion and (2) generation of the re- sponse that elicits a prespecified emotion in the ad- dressee’s mind.2 In the prediction task, the system is provided with a dialogue history. For simplic- ity, we consider, as a history, an utterance and a response to it (Figure 1). Given the history, the system predicts the addressee’s emotion that will be caused by the response. For example, the sys- tem …

Content truncated to 500 words