Towards a Learning Analytics Dashboard for Collaborative Conversational Agent Activities in MOOCs

Abstract

This paper presents the design of a learning analytics dashboard, utilizing learning traces that emerge from conversational MOOC activities. These chat-based activities encourage students to collaborate in dyads, with the support of a conversational agent, in order to answer open-ended questions, set by the course instructor. Taking into account the dynamic and context-sensitive nature of the agent-based support provided during this kind of online activities, we aspire to share a series of variables, metrics and use cases, which can be valuable for practitioners and researchers looking forward to initiate the development of a learning analytics dashboard that encompasses aspects of both peer-to-peer and human-to-bot conversational interactions.