Design and Implementation of the Adaptive Teachable Agent

Abstract

Traditionally intelligent tutoring system (ITS) uses expert model todeliver knowledge and train skills to users. According to the theoriesof leaning, learning by teaching method is more efficient for enhancingmotivation to learn and cognitive ability than learning by listening orlearning by reading. For the purpose of developing an intelligent agentto enhance the motivation to learn, the new type of teachable agent weredesigned and implemented, in which the user plays a role of a tutor byteaching the agent. In addition, we provide adaptive user interfacewhere each user plays his/her own scenario according to level ofinterests and motivation. We design the program to teach the agent about`rock cycle'. The program consists of four modules: teach module, Q&Amodule, test module, and network module. In teach module, the userteaches the agent and the agent's knowledge is structured and organized.In Q&A module, the user answer the question through an interactivewindow. In network module, the database server gathers log data of usersto measure users' interest and motivation about TA. It is expected thatproviding the user with the active role of teaching the agent enhancethe motivation to learn and the positive attitude toward the subjectmatter as well as cognition.