E-Learn: World Conference on E-Learning in Corporate, Government, Healthcare, and Higher Education, in Vancouver, British Columbia, Canada ISBN 978-1-939797-31-5 Publisher: Association for the Advancement of Computing in Education (AACE), San Diego, CA
Researchers have acknowledged the important role of conversational agents in education but there has been very little research on the types of user conversations and whether different types of users might require different types of conversational strategies to remain engaged throughout the interaction. In this paper, previous conversations with a conversational agent were examined in order to identify the different types of user styles and develop a machine learning algorithm that could perform the classification with new users. Moreover, an intervention designed to improve overall engagement was implemented for each classification and assessed with a new group of users (N=56). The results indicated that the classification was generally successful but the intervention only had a significant effect for one of the three conversational styles. These findings provide evidence for the importance of personalizing interactions with conversational agents so that engagement is maximized and conversational objectives are achieved.
Procter, M., Heller, B. & Lin, O. (2017). Assessing the impact of intelligent interventions by conversational agents: Implications for pedagogical agent design. In J. Dron & S. Mishra (Eds.), Proceedings of E-Learn: World Conference on E-Learning in Corporate, Government, Healthcare, and Higher Education (pp. 649-653). Vancouver, British Columbia, Canada: Association for the Advancement of Computing in Education (AACE).
© 2017 Association for the Advancement of Computing in Education (AACE)