INSPIRED: Toward Sociable Recommendation Dialog Systems

Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing, pages 8142–8152, November 16–20, 2020. c⃝2020 Association for Computational Linguistics 8142 INSPIRED: Toward Sociable Recommendation Dialog Systems Shirley Anugrah Hayati♡ Dongyeop Kang☆ Qingxiaoyang Zhu♡ Weiyan Shi♡ Zhou Yu♡ ♡Department of Computer Science, University of California, Davis ☆University of California, Berkeley {sahayati, qinzhu, wyshi, joyu}@ucdavis.edu dongyeopk@berkeley.edu Abstract In recommendation dialogs, humans com- monly disclose their preference and make rec- ommendations in a friendly manner. However, this is a challenge in developing a sociable rec- ommendation dialog system, due to the lack of dialog dataset annotated with such sociable strategies. Therefore, we present INSPIRED, a new dataset of 1,001 human-human dialogs for movie recommendation with measures for successful recommendations. To better under- stand how humans make recommendations in communication, we design an annotation scheme related to recommendation strategies based on social science theories and annotate these dialogs. Our analysis shows that sociable recommendation strategies, such as sharing personal opinions or communicating with en- couragement, more frequently lead to success- ful recommendations. Based on our dataset, we train end-to-end recommendation dialog systems with and without our strategy labels. In both automatic and human evaluation, our model with strategy incorporation outperforms the baseline model. This work is a first step for building sociable recommendation dialog sys- tems with a basis of social science theories1. 1 Introduction Sociable conversational agents build rapport with users, in order to gain trust and favor from them. So- cial science researchers believe that the rapport in- fluence a more persuasive recommendation to suc- cessfully suggest an item that satisfies user needs (Yoo et al., 2012; Gkika and Lekakos; Pecune et al., 2019; Gretzel and Fesenmaier, 2006). However, existing works on recommendation di- alog systems lack a study about communication strategies used by human speakers for making suc- cessful and persuasive recommendations. They col- 1Dataset and code are available at https://github. com/sweetpeach/Inspired REC: Hi! Happy Thanksgiving! I'm here to help you find a trailer! SEEK: Happy Thanksgiving! My favorite movie is finding Nemo I really like it REC: Awesome! So do you like Disney movies in general? SEEK: Yup they are so colorful and full of life! REC: Yeah, I love Disney too! I have Disney + and watch it everyday haha. Have you seen the new Lady and the Tramp? I find it relatable to my dog! SEEK: Lol that’s good enough! Never heard of that one! what is it about? REC: It's about a dog named Lady who runs away with a stray named Tramp out of jealousy . . What do you think? SEEK: Woo sounds good! I definitely want to see this. Thank you! REC: No problem! Hope you enjoy it as I did! OFFERING HELP ENCOURAGEMENT PERSONAL EXPERIENCE PREFERENCE CONFIRMATION PERSONAL OPINION SIMILARITY CREDIBILITY EXPERIENCE INQUIRY OPINION INQUIRY Figure 1: An example snippet of human-human recom- mendation dialog in INSPIRED. REC refers a person who recommends a movie and SEEK refers a person who looks for a recommendation. Above each recom- mender’s utterance …

Content truncated to 500 words