Icebreaking: Building Trust and Empathy with Virtual Museum Embodied Conversational Agents via Personalized Initial Interactions

Embodied Conversational Agents (ECAs) show significant potential for enhancing user experience in virtual museums. However, effectively establishing user trust and empathy during the critical initial moments of encounter remains a key challenge. Existing research has paid limited attention to systematically fostering these qualities through personalized Icebreaking strategies, often focusing instead on agent appearance or guidance mechanisms. This paper investigates the impact of a multimodal personalized initial interaction strategy—integrating personalized openings based on user interests, guided self-disclosure, and adaptive embodied social rituals—on users' trust in and empathy towards an ECA guide. We implemented this strategy within a VR virtual museum themed around the Beijing Central Axis, featuring interactions led by an ECA guide named “Xiao Liang.” In a user study ($N=60$), we compared three conditions: (1) No initial interaction (control), (2) Standardized interaction, and (3) Personalized interaction. User trust and empathy were measured immediately post-interaction using established psychological scales. Results clearly demonstrated that the personalized initial interaction significantly enhanced both user trust ($p<0.05$) and empathy ($p<0.05$) compared to both the control and standardized interaction groups. These findings confirm that personalizing the integration of dialogue and embodied behaviors early in the interaction is an effective pathway for building positive user-agent relationships. Furthermore, this work provides valuable insights for designing more trustworthy and empathic initial interactions for ECAs, offering a significant contribution to existing ECA interaction paradigms.