图书情报知识 ›› 2023, Vol. 40 ›› Issue (2): 29-39.doi: 10.13366/j.dik.2023.02.029

• 专题·人智交互体验度量与智能产品设计 • 上一篇    下一篇

社交机器人如何提升用户社会临场感? 表情包情感效价在人智对话交互中的作用

邓俊 1, 易欣妍 1, 傅诗婷 2   

  1. 1.武汉大学城市设计学院,武汉,430072;
    2.武汉大学信息管理学院,武汉,430072
  • 出版日期:2023-03-10 发布日期:2023-05-09
  • 通讯作者: 傅诗婷(ORCID:0000-0002-5636-7931),博士研究生,研究方向:人智交互、信息行为,Email:fushiting@whu.edu.cn。
  • 作者简介:邓俊(ORCID:0000-0001-6021-1668),博士,副教授,研究方向:智能产品设计,Email:whuteach@126.com;易欣妍(ORCID:0000-0003-0669-8369),硕士研究生,研究方向:智能产品设计,Email:yxy17259@outlook.com。
  • 基金资助:
    本文系国家社会科学基金重大项目“人本人工智能驱动的信息服务体系重构与应用研究”(22&ZD325)和武汉大学“351人才计划”教学岗位建设项目(珞珈青年学者)(WDBZ202234)的研究成果之一。

How to Improve Users Social Presence by Social Robots? The Role of Sticker’s Emotional Valence in Human-AI Conversation Interaction

DENG Jun, YI Xinyan, FU Shiting   

  • Online:2023-03-10 Published:2023-05-09
  • Contact: Correspondence should be addressed to FU Shiting, Email: fushiting@whu.edu.cn, ORCID: 0000-0002-5636-7931
  • Supported by:
    This is an outcome of the Major Project "Information Service System Restructuring and Application Driven by Human-Centered Artificial Intelligence" (22&ZD325)supported by National Social Science Foundation of China and the Teaching Post Construction Project(WDBZ202234)supported by a grant from "351 Talent Program" for Luojia Young Scholars of Wuhan University.

摘要: [ 目的 / 意义 ] 在人智交互过程中,聊天机器人可以使用表情包传达文字不具备的社交线索以提升用户体验。揭示聊天机器人的表情包与文本情感效价一致性对用户社会临场感及社会化反应的影响,以提高社交机器人的拟人化程度,可促进未来人们对机器人服务的广泛接受与使用。[ 研究设计 / 方法 ] 构建人智交互受控实验环境,采取2(表情包情感效价:积极 / 消极)x2(文本情感效价:积极 / 消极)组内设计,共招募45名参与者完成与聊天机器人的4轮对话任务,通过问卷调查数据度量表情包与文本效价一致组与不一致组中的用户社会临场感与社会化反应。[ 结论 / 发现 ] 在人智交互过程中,当聊天机器人使用情感效价一致的表情包与文本时,用户的社会临场感比不一致组更强,并作出更积极的社会化反应,具体表现为更愿意接受聊天机器人的建议并对其产生更高的信任度、满意度和使用意愿。[ 创新 / 价值 ] 增强了视觉 - 语言多模态人智交互体验研究的理论基础,为优化人智交互设计决策流程作出了方法论贡献,并为社交机器人的对话形式设计提供了实践指导。

关键词: 人智交互, 社交机器人, 表情包, 情感效价, 社会临场感, 社会化反应

Abstract: [Purpose/Significance] In the process of human-AI interaction, chatbots can enhance the user experience by using stickers to express the social cues that text can't convey during human-AI conversations. This study aims to reveal the impact of the congruence of the emotional valence of the sticker and the text on users' social presences and social responses. It will improve the anthropomorphism of social robots and facilitate the widespread acceptance and application of robotic services in the future. [Design/Methodology] The study created a simulated human-AI interaction scenario, meanwhile conducted a controlled experiment in a 2(sticker valence: positive/negative)×2(text valence: positive/negative)within-subject design, in which 45 participants were recruited to complete four rounds of conversational tasks with chatbots and then measured their social presences and social responses of users in congruent group and incongruent group of emotional valence of the sticker and the text through questionnaires. [Findings/Conclusion] Compared with the incongruent group, participants in the congruent group experienced stronger social presences during human-AI interactions and performed more positive social responses. Specifically, they were more willing to accept the chatbot's advice as well as had higher trust, satisfaction, and intention to use them. [Originality/Value] It enhances the theoretical foundation of experience studying in human-AI interaction under visual-verbal multi-modal interfaces, makes methodological contributions for human-AI interaction design decisions, and provides practical guidance for the dialogue design of social robots.

Keywords: Human-AI interaction, Social robots, Stickers, Emotional valence, Social presence, Social response