Experiences with BDI-based design and implementation of Intelligent Pedagogical Agents

Abstract

Intelligent Pedagogical Agents (IPAs) are intelligent agents that are designed for pedagogical purposes to support learning. Several benefits of IPAs have been found adding to support learning effectiveness. Pedagogical agents can be thought of as a central point of interaction between the learner and the environment. And hence, the intelligent behavior and functional richness of pedagogical agents have the potential to reward back into learning results and effectiveness. However, the realization of those agents remains to be a challenge based on intelligent agents. Intelligent agents' research contributes interesting behavior to pedagogical agents such as pro-activeness, re-activeness and communication with other agents and more. Designing intelligent agents for pedagogical purposes entails action reasoning that makes agents smart in selecting best courses of action (plans) to reach the pedagogical goals with the learner. Towards the objective of efficient design and implementation of intelligent pedagogical agents, we selected practical agent frameworks for evaluation such as JADE, JACK, 2APL, GOAL, JADEX, Jason, and others. We compare those platforms and explain a learning scenario that involves pedagogical agents aiding a learner to perform a physics experiment in a virtual environment. To that end, the paper exhibits the difficulties, resolutions, and decisions made when designing and implementing an intelligent BDI-based system, the design choices and the selected process are shown.