Interviews in social and health sciences are resource intensive and susceptible to interviewer bias, inconsistency, and variability across interviewers. Moreover, human-led interviews may inhibit participant openness, especially regarding sensitive topics, due to judgment, compromised anonymity, or discomfort in face-To-face interactions. These shortcomings limit the quality of the data collected. To this end, we propose the Embodied Conversational Interview Agentic Service (ELIAS). Informed by human-developed interview guides, ELIAS aims to streamline the interview process by combining an empathetic and bias-free embodied conversational interview agent with a semi-supervised content analysis and coding agent. We describe the development of the first version of ELIAS and also present results from a first evaluation study with five participants. We assessed the acceptance of and alliance with the embodied conversational interview agent. The evaluation shows positive perceptions and a strong alliance with the conversational agent. Suggestions for improvement will guide our future work.

UMAP Adjunct '25: Adjunct Proceedings of the 33rd ACM Conference on User Modeling, Adaptation and Personalization

ISBN:979-8-4007-1399-6