Automated Detection of Mentors and Players in an Educational Game

Automatically identifying the various roles (e.g., mentor, player) in multi-party collaborative chat is a challenging task. To better understand the conversational demands of mentors and players, this paper investigates the dynamics and linguistic features of multi-party chat in the context of an online educational game. In this paper we introduce a novel computational linguistics method using a machine learning algorithm to automatically classify utterances of players and mentors in a serious game, where players act as interns in an urban planning firm and discuss their ideas about urban planning and environmental science in written natural language. Our results are promising and our model can be extended to any multi-party environment that leaders (Mentors) are needed to be distinguished based on their conversation.