[2O6-OS-2b] 人間と共生する対話知能

[2O6-OS-2b-01] Spoken Dialogue System to overcome Dialogue Breakdown based on user’s Personality Traits

〇Mai Okada¹, Takuya Takeda¹, Kazuya Tsubokura¹, Yurie Iribe¹, Norihide Kitaoka² (1. Aichi Prefectural University, 2. Toyohashi University of Technology)

Keywords:Dialogue system, Dialogue breakdown, Personality traits, Multimodality

Automated Dialog systems have come to be used in various fields. However, it is necessary to detect and fix the dialog breakdown in the present dialog system since it has occurred. There was a problem that the breakdown recovery method proposed until now is one pattern. Therefore, in this study, we propose a new breakdown recovery method according to the user's dialog act and personality traits. First, we gathered the recovery utterances of the system for the breakdown from cloud workers, and we constructed a breakdown recovery corpus. Next, we analyzed the corpus and examined how the recovery utterance should be when the system is broken. Finally, it is clarified that the proposed method is effective due to implementing and evaluating the recovery method obtained by analyzing the system.