Dialogue Scenario Collection of Persuasive Dialogue with Emotional Expressions via Crowdsourcing

Dialogue Scenario Collection of Persuasive Dialogue with Emotional Expressions via Crowdsourcing Koichiro Yoshino1,2, Yoko Ishikawa1, Masahiro Mizukami1, Yu Suzuki1, Sakti Sakriani1, and Satoshi Nakamura1 1Graduate School of Information Science, Nara Institute of Science and Technology, 8916-5, Takayama-cho, Ikoma, 6300192, Japan 2PRESTO, Japan Science and Technology Agency, 4-1-8 Hon-machi, Kawaguchi, Saitama 332-0012, Japan {koichiro, ishikawa.yoko.io5, masahiro-mi, ssakti, ysuzuki, s-nakamura}@is.naist.jp Abstract Existing dialogue data collection methods such as the Wizard of Oz method (WoZ) or real dialogue recording are costly, and they prevent launching a new dialogue system. In this study, we requested crowd workers in crowdsourcing to create dialogue scenarios according to the instruction of the situation for persuasive dialogue systems that use emotional expressions. We collected 200 dialogues in 5 scenarios for a total of 1,000 via crowdsourcing. We also annotated emotional states and users’ acceptance for system persuasion by using crowdsourcing. We constructed a persuasive dialogue system with the collected data and evaluated the system by interacting with crowd works. From the experiment, it was investigated that the collected labels have sufficient agreement even if we did not impose any training of annotation to workers. Keywords: Dialogue corpus, emotion labels, persuasive dialogue, crowdsourcing 1. Introduction Expressing one’s emotion or feeling with language helps us to understand each other in human-human communica- tion. We cannot observe the internal states of other peo- ple such as our emotions or feelings directly, and we in- directly estimate the internal states from observable states such as utterances and the behaviors of others. Explicitly expressing internal states helps to build a good relationship between humans by reducing misunderstandings and un- requited feelings (Ekman, 1993). For example, sharing the same emotional state helps us to develop close relationships with others. Expressing emotional states is very effective to process the persuasion or negotiation (Keltner and Haidt, 1999; Mor- ris and Keltner, 2000). Persuaders who express positive emotions increase the ratio of success at persuasion more than persuaders who do not express their emotions because expressing positive emotions gives a cooperative impres- sion to a partner (Carnevale and Isen, 1986; Forgas, 1998). On the other hand, expressing negative emotions such as “anger” may wrest a concession from a partner even if the proposal from the persuader is not attractive to the partner, especially when the partner does not have any other options to choose (Sinaceur and Tiedens, 2006). Using emotional expressions is one of the most effective ways to change the belief or behaviors of a partner to increase the success rate of persuasion or negotiation. Captology (Computers as persuasive technologies) (Fogg, 1997) is a research area to work on systems that affect the beliefs or behaviors of users. A persuasive dialogue system is known as a part of this research area, and some dialogue systems have been developed that can change the actions or behaviors of users with persuasion. Some studies of persua- sive dialogue systems investigated efficient dialogue strat- egy on dialogue management to persuade users. Hiraoka et al., (2016) introduced actions of framing and …

Content truncated to 500 words