Effectiveness Analysis of a Method for Generating Work Reactions to Spoken Utterances for Facilitating Task Execution Using Haiku Writing

Abstract

Recently, a genre of streaming known as working streaming has emerged on video-streaming platforms like YouTube. Although working streaming has the potential to improve work productivity through self-explanation and viewer reactions, it faces challenges, such as confidentiality and the need for a significant number of viewers. To address these challenges, we previously proposed a new method to simulate real-time interaction by automatically generating viewer comments in response to user utterances. However, the experiment was conducted with four undergraduate students, leaving room for further verification. In this study, we conducted a large-scale experiment using crowdsourcing to verify the effectiveness of the proposed method.