The Collaborative System with Situated Agents for Activating Observation Learning

Abstract

This study focuses on modality of externalized knowledge-acquisition on problem solving via “Learning by Observing” and “Learning by Teaching/Explaining”. Among the learning effects expected by collaborative learning, we can suppose meta-cognition/distributed cognition such as reflective thinking, self-monitoring and so on. We propose the architecture of the intelligent collaborative system with situated agents for activating observation learning, which is a kind of virtual collaborative learning environment. The situated agents are embedded in this system, who have the different roles to make a real student activate observation learning. Each of the situated agents behaves dynamically based on the “Student Model” which represents the understanding state of a real student, and “Transforming Process Model” which stands for the converting process from internal state into external state for a level of acquired knowledge. A real student is expected to acquire any meta-knowledge via the interaction among situated agents in the collaborative learning environment. A facilitator agent who is a kind of a situated agent manages these Models and controls the other situated agents’ behavior. Moreover, it can identifies a real student’s bugs/faults occurred in the converting operations.