[4Xin1] Poster session 2

[4Xin1-67] Investigation on Personalization of Tone in Explaining Recommendation Results

〇Ayano Okoso¹, Keisuke Otaki¹, Satoshi Koide¹, Yukino Baba² (1.Toyota Central R&D Labs., Inc, 2.The University of Tokyo)

Keywords:recommender system, personalized explanation, justification

Acceptability in recommender systems is an important issue. Recently, research on explainable recommendations has been attracting attention to improve acceptability. In particular, we focus on "persuasiveness," which encourages users to consume items. A persuasive explanation can consist of the content of the explanation and how it is expressed. Personalization of these elements is expected to improve persuasiveness. However, while there are many previous studies on personalization of the content, to the best of our knowledge, there are few previous studies on personalization of the way of expression. Therefore, we focus on personalizing the way of expression in explaining recommendation results. Here, we refer to the way of expression as tone (e.g., emotional). This paper reports the results of a crowdsourcing experiment that revealed that tones that feel persuasive differ by individuals.